21 Novembre 2026

Psicologia Punto One | Psicologia e Psicologi nel Web

Menu

Menu

SCHEDA SCIENTIFICA2026Altra ricerca

Diabetes mellitus is associated with altered Alzheimer's disease pathology.

Min Wu, Ying Luan, Xiaowen Xu, Xiaoxie Mao, Huiying Lin, Kun He, Qi Huang, Jie Zhang, Sheng Kuang, Yihui Guan, Alzheimer's Disease Neuroimaging Initiative, Health and Aging Brain Study (HABS‐HD) Team, Sid O'Bryant, Binyin Li, Qihao Guo, Fang Xie, Benyan Luo

Alzheimer's & dementia : the journal of the Alzheimer's Association · 2026

Abstract originale

INTRODUCTION: Diabetes mellitus (DM) is a major risk factor for dementia, yet it remains unclear whether DM accelerates cognitive decline through amyloid beta (Aβ) or tau accumulation in Alzheimer's disease (AD). METHODS: In this study, we included 2784 participants from three independent observational cohorts: the Chinese Preclinical Alzheimer's Disease Study (CPAS), the Alzheimer's Disease Neuroimaging Initiative (ADNI), and Health and Aging Brain Study-Health Disparities (HABS-HD). We examined the associations of DM with Aβ positron emission tomography (PET), tau PET, plasma phosphorylated tau (p-tau)217, and longitudinal cognitive change. We also tested whether DM modified the associations of Aβ and plasma p-tau217 with subsequent tau accumulation and explored the association of metformin use with tau accumulation. RESULTS: We found that DM was associated with lower global Aβ burden and lower tau accumulation in CPAS and ADNI, in which cognitive impairment was more common than in HABS-HD. The associations of Aβ and plasma p-tau217 with tau accumulation were weaker among individuals with DM. In addition, DM was associated with reduced propagation of tau from early to later stage Braak regions. Longitudinal analyses in ADNI indicated that individuals with DM exhibited slower tau-related cognitive decline. In exploratory analyses, metformin use among individuals with DM was associated with lower tau accumulation compared to individuals without DM. DISCUSSION: These findings suggest that DM is associated with an attenuation of amyloid- and tau-related processes, particularly in participants with cognitive impairment, and further suggest that diabetes-related mechanisms and its management may modulate the progression of AD pathology.

Parole chiave

HumansAlzheimer DiseaseFemaletau ProteinsPositron-Emission TomographyAmyloid beta-PeptidesMaleAgedBrainDiabetes MellitusCognitive DysfunctionAged, 80 and overLongitudinal StudiesAlzheimer's diseaseamyloid betadementiadiabetes mellitustau accumulation
Leggi il testo integrale originale

Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.

Diabetes mellitus is associated with altered Alzheimer's disease pathology WU et al.

Abstract

INTRODUCTION

Diabetes mellitus (DM) is a major risk factor for dementia, yet it remains unclear whether DM accelerates cognitive decline through amyloid beta (Aβ) or tau accumulation in Alzheimer's disease (AD).

METHODS

In this study, we included 2784 participants from three independent observational cohorts: the Chinese Preclinical Alzheimer's Disease Study (CPAS), the Alzheimer's Disease Neuroimaging Initiative (ADNI), and Health and Aging Brain Study–Health Disparities (HABS‐HD). We examined the associations of DM with Aβ positron emission tomography (PET), tau PET, plasma phosphorylated tau (p‐tau)217, and longitudinal cognitive change. We also tested whether DM modified the associations of Aβ and plasma p‐tau217 with subsequent tau accumulation and explored the association of metformin use with tau accumulation.

RESULTS

We found that DM was associated with lower global Aβ burden and lower tau accumulation in CPAS and ADNI, in which cognitive impairment was more common than in HABS‐HD. The associations of Aβ and plasma p‐tau217 with tau accumulation were weaker among individuals with DM. In addition, DM was associated with reduced propagation of tau from early to later stage Braak regions. Longitudinal analyses in ADNI indicated that individuals with DM exhibited slower tau‐related cognitive decline. In exploratory analyses, metformin use among individuals with DM was associated with lower tau accumulation compared to individuals without DM.

DISCUSSION

These findings suggest that DM is associated with an attenuation of amyloid‐ and tau‐related processes, particularly in participants with cognitive impairment, and further suggest that diabetes‐related mechanisms and its management may modulate the progression of AD pathology.

Highlights

Diabetes mellitus was associated with lower global amyloid beta (Aβ) burden and slower tau accumulation.

Associations of Aβ and plasma phosphorylated tau 217 with tau accumulation were weaker among individuals with diabetes mellitus.

Diabetes mellitus was associated with reduced tau propagation from early to later Braak regions.

Individuals with diabetes mellitus showed slower tau‐related cognitive decline.

1INTRODUCTION

Diabetes mellitus (DM) affects approximately one in ten adults with prevalence projected to rise to 1.31 billion by 2050.1 DM is a risk factor for cognitive decline and dementia.2, 3, 4 Large epidemiologic studies consistently demonstrate that DM nearly doubles the risk of dementia 5, 6, 7. However, the effect of DM on cognitive trajectories was inconsistent, while most investigations have assumed DM accelerates Alzheimer's disease (AD) pathology and cognitive decline. Some longitudinal studies indicate that DM is associated with an increased risk of progression to AD and with faster cognitive decline 8, 9, whereas others have observed DM is linked to slower or unchanged rate of cognitive decline among patients with established AD 10, 11. This heterogeneity raises a question of how DM contributes to dementia and whether it acts through AD‐related neuropathology.

AD is characterized by the accumulation of amyloid beta (Aβ) plaques and the aggregation and spread of pathological tau neurofibrillary tangles 12, 13, 14. The extent to which DM modifies these core AD pathologies remains controversial. Some studies report greater Aβ burden on positron emission tomography (PET) in individuals with DM 15, while others find no clear association 16, 17. Tau pathology has been examined less extensively, and findings are controversial. While earlier studies reported higher cerebrospinal fluid (CSF) total and phosphorylated tau (p‐tau) in DM 16, recent multicenter data indicate that DM is associated with lower CSF p‐tau, with notable heterogeneity between population‐based and memory clinic cohorts 18. Taken together, existing evidence challenges the notion that DM uniformly exacerbates typical AD pathology.

Blood biomarkers, particularly p‐tau217, have emerged as cost‐effective, easily accessible, and highly sensitive tools to detect and monitor AD pathology 19, 20. Plasma p‐tau217 shows high accuracy for identifying abnormal brain Aβ and tau burdens even in a preclinical stage and is thought to be related to both the Aβ and tau pathology 21, 22. However, it remains unclear whether plasma p‐tau217 differs by DM status and whether DM modifies the relationships of Aβ burden and plasma biomarkers with tau pathology and cognitive decline.

To address these gaps, we examined Aβ and tau PET alongside plasma tau biomarkers in participants from three independent multicenter cohorts in 2784 participants with both Aβ and tau PET scans. We compared Aβ and tau burden, and plasma p‐tau217 levels between individuals with and without diabetes. We then tested whether DM modifies (1) the association between plasma p‐tau217 levels and tau accumulation, (2) the association between Aβ burden and tau accumulation, and (3) the spatial propagation of tau from early stage regions to later stage Braak regions. Finally, we evaluated whether DM alters the trajectory of tau‐associated longitudinal cognitive decline.

2METHODS

2.1Participants

RESEARCH IN CONTEXT

Systematic review: We searched PubMed for studies examining diabetes in relation to cognitive decline, amyloid beta (Aβ), tau, or Alzheimer's disease (AD) biomarkers. Epidemiological evidence consistently links diabetes to dementia risk, whereas autopsy, fluid biomarker, and positron emission tomography (PET) studies have reported absent, inconsistent, or inverse associations with AD pathology. Few studies have jointly evaluated longitudinal amyloid PET, tau PET, plasma phosphorylated tau (p‐tau)217, regional tau propagation, and cognition across independent cohorts.

Interpretation: We found that diabetes was associated with lower Aβ burden, slower tau accumulation and propagation, attenuation of Aβ– and p‐tau217–related tau accumulation, and slower tau‐related cognitive decline, particularly in cognitively impaired participants. Exploratory analyses reported that metformin use among individuals with diabetes mellitus (DM) was associated with lower tau accumulation compared to individuals without DM.

Future directions: Prospective studies should determine whether these associations reflect diabetes biology, treatment, or competing vascular pathology, while incorporating diabetes duration, glycemic control, vascular burden, medication exposure, and repeated AD biomarkers.

A total of 2784 participants were included from three independent cohorts: the Chinese Preclinical Alzheimer's Disease Study (CPAS),23 Alzheimer's Disease Neuroimaging Initiative (ADNI), and the Health and Aging Brain Study–Health Disparities (HABS‐HD). Study protocols were approved by the institutional review boards of all participating institutions and written informed consent was obtained from all participants.

CPAS is an ongoing prospective, multicenter study initiated in 2019 to characterize multimodal biomarkers of AD in a Chinese population, with recruitment from memory clinics at three hospitals in Shanghai 23. Detailed inclusion and exclusion criteria for this cohort have been described previously 24, 25. ADNI is a longitudinal, multicenter study launched in 2003 across North America to track AD progression. HABS‐HD is a community‐based longitudinal study designed to investigate biomarkers of AD and related dementias in diverse older adults.

In CPAS, dementia was diagnosed according to the 2018 National Institute on Aging–Alzheimer's Association (NIA‐AA) criteria and mild cognitive impairment (MCI) was defined using a modified Jak–Bondi neuropsychological test across three cognitive domains, that is, episodic memory, language function, and executive function 26, 27. In ADNI, diagnoses followed established ADNI Clinical Core criteria 28, and in HABS‐HD diagnoses were made by decision‐tree algorithm followed by expert consensus review.29 Participants with MCI or dementia were grouped as cognitively impaired (CI); all others without cognitive impairment were classified as cognitively unimpaired (CU).

In CPAS and HABS‐HD, DM was identified based on medical history obtained through structured interviews. In ADNI, DM status was determined through medical history databases and medication records (documented use of insulin or hypoglycemics). Participants were therefore classified as DM or non‐DM. Eligibility criteria included a documented diagnosis of DM or confirmed absence of DM, availability of both Aβ and tau PET scans acquired within 6 months of each other, and availability of 3T magnetic resonance imaging (MRI).

2.2Plasma p‐tau 217

Plasma samples were collected from participants concurrently with PET scan (within 6 months). Plasma p‐tau217 was available in a subset of participants across cohorts. In CPAS, plasma p‐tau217 was measured using the commercial ALZpath p‐tau217 immunoassay (ALZpath, Inc). 20, 30 In ADNI, plasma p‐tau217 was measured using the same ALZpath assay on the Quanterix Simoa HD‐X platform at the Quanterix Accelerator Laboratory, with protocols available through the ADNI database. In HABS‐HD, plasma p‐tau217 was measured on the Quanterix Simoa platform.

2.3Neuroimaging

In CPAS, Aβ and tau PET were conducted on a PET/computed tomography scanner (Biograph‐mCT Flow, Siemens). Aβ PET used 18F‐florbetapir (FBP, ≈ 10 mCi) with a 20‐minute emission acquisition from 50 to 70 minutes post‐injection. Tau PET used 18F‐MK‐6240 (185 MBq ± 10%) with a 20‐minute emission acquisition from 90 to 110 minutes post‐injection. High‐resolution T1‐weighted structural MRI was acquired on a 3T Siemens Prisma scanner using a magnetization‐prepared rapid gradient‐echo (MPRAGE) sequence (repetition time = 3000 ms; echo time = 2.56 ms; flip angle = 7°; matrix = 320 × 320; voxel size = 0.8 mm isotropic; 208 sagittal slices). PET images were reconstructed using filtered back‐projection with corrections for attenuation, normalization, scatter, and random coincidences, resulting in a final voxel size of 2.04 × 2.04 × 1.50 mm3.

In ADNI, Aβ PET used 18F‐florbetapir (≈ 10 mCi), with emission data collected as four 5‐minute frames from 50 to 70 minutes post‐injection. Tau PET used 18F‐flortaucipir (FTP, ≈ 10 mCi), with emission data collected as six 5‐minute frames from 75 to 105 minutes post‐injection. Structural MRI was acquired on 3T Siemens, GE, or Philips scanners using MPRAGE sequences. Detailed acquisition and preprocessing procedures are available at http://adni.loni.usc.edu/methods/pet‐analysis‐method/.

In HABS‐HD, Aβ PET was acquired using 18F‐florbetaben (FBB, ≈ 8.1 mCi), with a 20‐minute acquisition from 90 to 110 minutes post‐injection. Tau PET used 18F‐PI‐2620 (≈ 10 mCi), with a 30‐minute acquisition from 45 to 75 minutes post‐injection. Structural MRI was acquired on 3T Siemens Prisma or GE Discovery scanners using MPRAGE sequences.

2.4Image processing

For CPAS and HABS‐HD, PET image preprocessing was performed using SPM12 (Wellcome Trust Centre for Neuroimaging, https://www.fil.ion.ucl.ac.uk/spm). PET images were co‐registered to each participant's T1‐weighted MRI, which were segmented into gray matter, white matter, and CSF. Spatial normalization to standard Montreal Neurological Institute (MNI) space was performed using deformation fields derived from T1‐weighted segmentation and subsequently applied to the co‐registered PET images.

For Aβ PET, regional uptake values were extracted using Centiloid standard mask images (voi_ cxt_2 mm.nii and voi_WhlCb_2 mm.nii) provided by the Global Alzheimer's Association Interactive Network (GAAIN). Standardized uptake value ratios (SUVRs) were calculated using whole cerebellum as the reference region and were conversed to the Centiloid value using tracer‐specific calibration equations (more details are available in our previous publication27). For tau PET, regional tau deposition was quantified for three composite regions corresponding to Braak stages I to II, III to IV, and V to VI using the Desikan–Killiany atlas. SUVRs were calculated using the inferior cerebellar gray matter as the reference region.

For ADNI, PET and MRI preprocessing and quantification followed the standardized pipelines as described by the ADNI PET Core and MRI Core, including motion correction, co‐registration to MRI, normalization, and reference region–based SUVR calculation. Amyloid PET SUVRs were expressed on the Centiloid scale, and tau PET SUVRs were derived using established ADNI reference regions. Detailed procedures are available on the ADNI website (http://ida.loni.usc.edu).

2.5Statistics

All statistical analyses were conducted using R (version 4.3.1) and MATLAB (R2023b). Normality of continuous variables was assessed using the Shapiro–Wilk test. Plasma biomarker concentrations were natural log transformed. Continuous variables are reported as mean ± standard deviation (SD), and categorical variables as counts and percentages. Participants’ characteristics were compared between groups stratified by cognitive status (i.e., CU and CI) and DM status using chi‐squared tests for categorical measures and analysis of variance for continuous measures.

Associations between DM and cross‐sectional measures (Aβ PET Centiloid, tau PET SUVR, plasma p‐tau217, and Mini‐Mental State Examination [MMSE]) were examined using linear mixed‐effects models. All models were adjusted for age, sex, education, cognitive status, and apolipoprotein E (APOE) ε4 carrier status. To account for multiple comparisons, false discovery rate (FDR) correction was applied.

The modification effect of DM was evaluated by including interactions (DM × plasma p‐tau217 and DM × Aβ PET Centiloid) in models predicting tau PET SUVR. To assess whether DM modified the association between early and later tau deposition, additional models were fitted in which tau accumulation in later Braak‐stage regions (Braak III–IV or Braak V–VI SUVR) was modeled as a function of the interaction between early‐stage tau accumulation (Braak I–II SUVR) and DM status.

Longitudinal analyses were conducted in the ADNI cohort. Individual MMSE slopes were estimated using linear regression over time. Linear mixed‐effects models were fitted to evaluate whether baseline tau PET SUVR, in interaction with DM status, predicted longitudinal change in MMSE, adjusting for the same covariates. To further characterize the relationships among DM, AD pathology, and longitudinal cognition change, structural equation modeling was performed using the lavaan package in R.

Finally, exploratory effects of antidiabetic treatments on AD pathology were evaluated using multivariable linear regression models adjusted for age, sex, education, cognitive status, and APOE ε4 carrier status, with post hoc pairwise comparisons between medication groups. All statistical tests were two sided, with a p value < 0.05 considered statistically significant.

3RESULTS

3.1Demographics

This study included 2784 participants from the CPAS (n = 734), ADNI (n = 835), and the HABS‐HD (n = 1215), all of whom underwent both Aβ PET and tau PET imaging (Table 1). Participants in ADNI were older than those in CPAS and HABS‐HD (mean [SD] age, 73.7 [7.9] vs. 68.1 [8.2] and 65.7 [8.5] years). The proportion of CI participants varied across cohorts, ranging from 27.9% in HABS‐HD to 44.7% in ADNI and 66.3% in CPAS. CI participants had lower education attainment, a higher prevalence of APOE ε4 carriers, and a greater proportion of Aβ positivity compared to CU participants. Female participants accounted for 453 (61.7%) in CPAS, 430 (51.5%) in ADNI, and 735 (60.5%) in HABS‐HD; diabetes prevalence was 15.7% in CPAS, 13.1% in ADNI, and 27.3% in HABS‐HD. Longitudinal cognitive assessments were available for a subset of 625 participants in the ADNI cohort over a mean (SD) follow‐up duration of 3.5 (1.7) years. Detailed baseline and longitudinal characteristics are reported in Table 1 and Table S1 in supporting information.

3.2Lower Aβ and tau accumulation in DM

We first compared Aβ and tau accumulation between participants with and without DM within each cohort. In CPAS and ADNI, participants with DM exhibited lower global Aβ burden and lower tau accumulation in temporoparietal regions (Figure 1A–B). After adjustment for age, sex, education, cognitive status, and APOE ε4 carrier status, diabetes was associated with lower Aβ Centiloids in CPAS (standardized β = −0.10; 95% confidence interval [95% CI], –0.17 to –0.04; p = 0.002, FDR corrected) and ADNI (β = −0.08; 95% CI, −0.14 to −0.03; p = 0.013, FDR corrected). DM was also associated with lower tau accumulation across Braak regions in both cohorts (all p < 0.05, FDR corrected). In HABS‐HD, no significant difference between DM and non‐DM was observed in either Aβ burden (β = −0.04; 95% CI, −0.09 to 0.01; p = 0.32) or tau accumulation (Braak I–II: β = 0.02; 95% CI, −0.04 to 0.07; p = 0.489; Braak III–IV: β = 0.05; 95% CI, −0.01 to 0.10; p = 0.277; Braak V–VI: β = 0.06; 95% CI, 0.01 to 0.12; p = 0.194, FDR corrected).

When stratified by cognitive status, that is, CU and CI, the associations of diabetes with lower Aβ and tau accumulation were primarily observed in CI participants in CPAS and ADNI (Figure S1 in supporting information). These associations remained consistent after additional adjustment for baseline MMSE scores. Among CU participants, associations were weaker and heterogeneous across cohorts: diabetes was associated with higher late‐stage tau in HABS‐HD, whereas associations were not consistently observed in CPAS or ADNI (Figure S1). Additional analyses stratified by Aβ status and analyses with further adjustment for white matter hyperintensity volume are presented in Figures S6 and S7 in supporting information, respectively. The associations of APOE ε4 carrier status with AD biomarkers within the DM and non‐DM groups are shown in Figure S8 in supporting information.

3.3Diabetes attenuated plasma p‐tau217– and Aβ–associated tau accumulation

We examined whether the association between plasma p‐tau217 and tau accumulation differed according to DM status. In linear mixed‐effects models adjusted for age, sex, education, APOE ε4 status, and cognitive status, higher plasma p‐tau217 was associated with higher tau accumulation across Braak regions in all three cohorts. However, this association was attenuated in participants with DM, as shown by significant DM × p‐tau217 interactions in CPAS (Braak I–II: standardized β = −0.24, 95% CI −0.42 to −0.05, p = 0.012; Braak III–IV: β = −0.26, 95% CI −0.46 to −0.07, p = 0.007; Braak V–VI: β = −0.32, 95% CI −0.53 to −0.12, p = 0.002; Figure 2A) and ADNI (Braak I: β = −0.34, 95% CI −0.64 to −0.05, p = 0.024; Braak III–IV: β = −0.47, 95% CI −0.77 to −0.16, p = 0.003; Braak V–VI: β = −0.47, 95% CI −0.78 to −0.16, p = 0.003; Figure 2B). In HABS‐HD, DM attenuated the association between p‐tau217 and tau PET SUVR in moderate‐ and late‐stage Braak regions (Braak III–IV: β = −0.53, 95% CI −0.82 to −0.24, p < 0.001; Braak V–VI: β = −0.56, 95% CI −0.86 to −0.26, p < 0.001), but not in Braak I and II (β = −0.14, 95% CI −0.44 to 0.17, p = 0.38; Figure 2C).

We next tested whether DM similarly modified the association between Aβ burden and tau accumulation. DM significantly attenuated the Aβ–tau association in late Braak stages in CPAS (Braak V–VI: β = −0.26, 95% CI −0.44 to −0.07, p = 0.006; Figure 3A) and in moderate Braak regions in ADNI (Braak III–IV: β = −0.22, 95% CI −0.43 to −0.01, p = 0.039; Figure 3B). No significant modification of the Aβ–tau association by DM was detected in HABS‐HD at any Braak stage (all p > 0.05; Figure 3C). Stratified analyses by CU and CI indicated that these attenuating effects of DM on p‐tau217–tau and Aβ–tau associations were primarily observed among CI participants (Figures S2–S3 in supporting information). In comparison, DM did not modify the glial fibrillary acidic protein (GFAP)‐related Aβ and tau burden as indicated by no significant interactions between DM and Aβ or tau accumulation (all p > 0.05, Figure S4 in supporting information).

3.4Diabetes attenuated tau propagation across Braak stages

We next examined whether the association between early‐ and late‐stage tau accumulation differed according to DM status by fitting linear mixed‐effects models including an interaction between DM and early‐stage tau (Braak I–II SUVR), adjusted for the same covariates. In CPAS, higher Braak I and II tau was less associated with Braak V and VI tau among participants with DM, as indicated by a significant DM × Braak I and II tau interaction (β = −0.24, 95% CI −0.40 to −0.08, p = 0.003; Figure 3D). In ADNI and HABS‐HD, DM attenuated the association between Braak I/I and II and Braak III and IV tau (ADNI: β = −0.22, 95% CI −0.35 to −0.08, p = 0.002; HABS‐HD: β = −0.14, 95% CI −0.23 to −0.04, p = 0.004; Figure 3E‐F).

3.5Diabetes moderated longitudinal tau‐related cognitive decline

We then assessed whether the association between baseline AD pathology and longitudinal cognitive decline differed according to DM status in ADNI. Significant interactions were observed between diabetes and baseline tau accumulation in models predicting longitudinal changes in the MMSE (Braak I: β = 0.573, p = 0.011; Braak III–IV: β = 0.672, p = 0.032; Braak V–VI: β = 0.957, p = 0.006; Figure 4A), indicating that for a given level of baseline tau accumulation, participants with DM exhibited slower tau‐related cognitive decline.

To further characterize the relationships among DM, AD pathology, and longitudinal cognitive change, structural equation modeling showed that DM was associated with lower Aβ burden (β = –0.06; 95% CI, −0.35 to −0.01, p = 0.03). Lower Aβ burden was, in turn, associated with reduced tau accumulation across all Braak regions; and both lower Aβ and tau burden were associated with a slower rate of MMSE decline (Figure 4B). Significant total indirect effects of DM on MMSE slope through Aβ and tau pathways were observed at Braak I (β = 0.08; 95% CI, 0.01 to 0.16; p = 0.048), Braak III and IV (β = 0.12; 95% CI, 0.04 to 0.22; p = 0.011), and Braak V and VI (β = 0.10; 95% CI, 0.02 to 0.18; p = 0.023).

3.6Exploratory associations of antidiabetic medication use with AD‐related pathology

Finally, we exploratorily examined whether antidiabetic medication use was associated with AD‐related pathology in CPAS and ADNI (medication details were unavailable in HABS‐HD). Compared to participants without DM, metformin users showed lower tau accumulation prominently in late‐stage regions (Braak V–VI) in CPAS (p = 0.023, Figure 5A) and across all Braak regions in ADNI (all p < 0.05, Figure 5B). Among participants with diabetes not receiving metformin, tau accumulation differed from that of non‐diabetic participants in ADNI at Braak III and IV (p = 0.041), with weaker evidence at Braak I (p = 0.062), and in CPAS at Braak V and VI (p = 0.099). Stratified analyses by CU and CI indicated that these associations were primarily present among CI participants and were not observed among CU individuals (Figure S5 in supporting information).

4DISCUSSION

In this multicenter study across three independent cohorts, we examined the associations of DM with Aβ and tau pathology using amyloid and tau PET, and plasma p‐tau217. Contrary to the hypothesis that DM exacerbates AD pathology, we found that DM was associated with lower Aβ burden and tau accumulation, particularly among CI participants. DM attenuated plasma p‐tau217‐ and amyloid‐related tau accumulation and weakened the association between early and later Braak stage tau accumulation. Longitudinally, DM also attenuated tau‐related cognitive decline, with slower decline observed at a given level of baseline tau accumulation. In exploratory analyses, metformin use among participants with DM was associated with lower tau accumulation. These findings suggest that DM, and potentially its treatment, may be associated with an altered relationship between AD biomarkers and cognitive progression.

We found that DM was associated with lower Aβ and tau burden, with these inverse associations being most pronounced among CI individuals. This finding is consistent with large‐scale neuropathological evidence. Autopsy studies have shown that individuals with DM do not exhibit increased global amyloid plaque or neurofibrillary tangle burden compared to non‐DM individuals, and in some studies, AD pathology is even less frequently observed in individuals with DM 31, 32, 33, 34, 35. However, the lower amyloid or tau burden in participants with DM should not be interpreted as evidence of neuroprotection. Instead, evidence suggests that DM is more strongly associated with cerebrovascular lesions and other non‐AD pathologies. Such vascular and metabolic brain injury may contribute to cognitive impairment in DM. Thus, the findings may reflect pathological substitution rather than protection against AD pathology. Consistent with this, amyloid PET studies report either no association or inverse associations between DM and fibrillar Aβ deposition 17, 36. However, in population‐based data, a diagnosis of diabetes ≈ 7 years prior to PET imaging has been linked to higher amyloid deposition 15, underscoring the heterogeneity and stage dependence of DM effects on AD pathology.

Our cohort‐level pattern supports this stage‐dependent interpretation. In CPAS and ADNI cohorts, which contain a higher proportion of CI individuals, DM was consistently associated with lower Aβ and tau burden, whereas in HABS‐HD, which is predominantly CU, we did not observe a comparable reduction in Aβ or tau; instead, DM was linked to higher late‐stage tau in CU participants. This stage‐dependent pattern is further supported by large‐scale biomarker evidence. In a meta‐analysis including > 5500 participants, DM was associated with lower odds of abnormal Aβ biomarkers in CI individuals, while higher odds of abnormal neurodegeneration were observed primarily among CU participants 18. However, individuals with DM who survived and were sufficiently healthy to enroll and undergo PET imaging may represent a selected subgroup. Other individuals with more severe metabolic or vascular disease, or a greater combined burden of vascular and AD pathology, may have died earlier or been less likely to enter or remain in these cohorts. Therefore, these findings should not be interpreted as evidence that DM is protective at later disease stages.

Beyond the lower overall Aβ and tau burden, diabetes was associated with attenuated Aβ‐related tau and plasma p‐tau217–related tau. This result suggests that tau accumulation in DM may be less linked to amyloid‐related pathways than in non‐diabetic individuals. Existing evidence shows that DM is strongly associated with cerebrovascular injury, metabolic dysregulation, and chronic inflammation, all of which can independently contribute to neurodegeneration and cognitive impairment 31, 33, 37, 38. Therefore, non‐amyloidogenic cascades may contribute more prominently to the pathology related to cognitive decline in DM individuals, thereby reducing the burden of classical amyloid–tau pathology required at the symptomatic stage.

Notably, our exploratory analyses did not identify evidence that diabetes modified the association between plasma GFAP (i.e., a marker of astrocytic activation) and Aβ or tau burden. It provides limited support for astrocytic activation (as indexed by GFAP) as the main contributor; however, this finding does not exclude the contribution of inflammation via other pathways. Another possible explanation for the lower Aβ‐ and p‐tau217–related tau accumulation observed in DM may be disease management. In our study, lower Aβ burden and tau accumulation were observed among metformin‐treated participants with DM than among participants without DM. Metformin use was associated with a lower tau accumulation, with weaker evidence for Aβ. This result is supported by preclinical studies suggesting that metformin modulates adenosine monophosphate activated protein kinase (AMPK) signaling and enhances protein phosphatase‐2A (PP2A) activity, thereby limiting tau phosphorylation 39, 40, promoting the clearance of pathological protein aggregates and limiting the spatial propagation of tau 41, 42. Although earlier animal studies raised concerns about potential amyloidogenic effects of metformin via beta‐secretase 1 upregulation 43, other experimental data, together with our findings, suggest a net neuroprotective effect and reduction of Aβ 41, 44. In addition to metformin, other glucose‐lowering therapies including glucagon‐like peptide‐1 receptor agonists and insulin, have been shown to exert neuroprotective effects that extend beyond glycemic control 45, 46, 47, 48, 49, 50. The direct comparisons between metformin users and non‐users with DM showed no consistent differences in Aβ or tau PET measures. The observed patterns may therefore relate more broadly to diabetes management, including glycemic control and the use of other glucose‐lowering medications.

Although diabetes is a risk factor for dementia, our longitudinal analyses in ADNI did not show faster cognitive decline among participants with diabetes. Instead, DM moderated the association between baseline tau accumulation and subsequent MMSE change, such that higher tau was less strongly coupled to cognitive decline in individuals with diabetes. Together with our cross‐sectional findings of lower Aβ and tau accumulation in diabetes in the more CI cohorts (i.e., CPAS and ADNI), these results support the notion that diabetes does not necessarily accelerate tau‐driven cognitive decline once substantial AD pathology is already present. Instead, DM may increase vulnerability to cognitive impairment through non‐AD mechanisms; or DM management may additionally modulate the AD‐related pathology and attenuate the rate of cognitive decline. This result aligns with a recent finding that DM was not associated with faster longitudinal brain Aβ deposition 51.

A key strength of this study is that it represents, to our knowledge, the largest multi‐cohort study to date examining the association between DM and AD‐related neuropathology in which all participants underwent both amyloid and tau PET (N = 2784), and with complementary plasma p‐tau217 measures available in a subset of participants. By analyzing data across three multicenter cohorts, we provide mechanistic insight into the interplay between DM and AD pathology. Several limitations should be acknowledged. First, detailed information on antidiabetic treatment exposure (e.g., duration, dosage), and differentiation between monotherapy and polypharmacy, was not available across cohorts; therefore, future studies should systematically investigate the effects of diabetes management on AD pathology. Second, longitudinal imaging data were not consistently available across cohorts, which constrained the assessment of long‐term Aβ and tau trajectories. Third, we did not differentiate type 1 and type 2 diabetes, precluding diabetes subtype‐specific analyses; although type 2 diabetes is much more frequent than type 1 in older people. Fourth, the duration of cognitive impairment and time since diagnosis were not consistently available across cohorts. Although the associations of DM with lower amyloid and tau burden remained significant after adjustment for baseline MMSE, residual confounding by disease duration or other aspects of clinical stage cannot be excluded.

5CONCLUSION

This multicohort study shows that diabetes mellitus is associated with lower tau pathology and an attenuated amyloid–tau association, reduced propagation of tau pathology, and slower tau‐related cognitive decline. Metformin use among individuals with DM was associated with lower tau accumulation. These findings suggest that diabetes, and potentially its management, may affect AD pathology, highlighting the need to develop diabetes‐specific management strategies for individuals with or at risk of dementia.

AUTHOR CONTRIBUTIONS

M.W.: conceptualization, methodology, data analysis, writing – original draft, writing – review and editing. Y.L.: conceptualization, methodology, data analysis, writing – review and editing. X.X.: methodology, data analysis, writing – review and editing; X.M.: conceptualization, data acquisition, writing – review and editing; H.L.: data acquisition, writing – review and editing; K.H.: data acquisition, writing – review and editing; Q.H.: methodology, writing – review and editing. J.Z.: methodology, writing – review and editing; S.K.: methodology, writing – review and editing; Y.G.: methodology, resources, funding acquisition, writing – review and editing; S.B.: resources, writing – review and editing; B.L.: data acquisition, resources, writing – review and editing; Q.G.: data acquisition, resources, writing – review and editing; B.Y.: conceptualization, supervision, resources, funding acquisition, writing – review and editing; F.X.: conceptualization, study design, methodology, supervision, resources, funding acquisition, writing – review and editing.

CONFLICT OF INTEREST STATEMENT

All authors declare that they have no conflicts of interest to disclose. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All human subjects provided informed consent.

Supporting information

REFERENCES

1Ong KL , Stafford LK , McLaughlin SA , et al. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021 . The Lancet . 2023 ;402 :203 ‐234 . doi:10.1016/S0140-6736(23)01301-6
PMC10364581 37356446

2Livingston G , Huntley J , Sommerlad A , et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission . The Lancet . 2020 ;396 :413 ‐446 . doi:10.1016/S0140-6736(20)30367-6
PMC7392084 32738937

3Livingston G , Sommerlad A , Orgeta V , et al. Dementia prevention, intervention, and care . The Lancet . 2017 ;390 :2673 ‐2734 . doi:10.1016/S0140-6736(17)31363-6
28735855

4Chen H , Dai H , Zhang Li , et al. The biomarker and clinical changes across the Alzheimer's continuum study (BCAS): rationale, design, and baseline characteristics of the first 1,013 participants . Alz Res Therapy . 2025 ;18 :18 . doi:10.1186/s13195-025-01937-x
PMC12837085 41437118

5Biessels GJ , Staekenborg S , Brunner E , Brayne C , Scheltens P . Risk of dementia in diabetes mellitus: a systematic review . Lancet Neurol . 2006 ;5 :64 ‐74 . doi:10.1016/S1474-4422(05)70284-2
16361024

6Cao F , Yang F , Li J , et al. The relationship between diabetes and the dementia risk: a meta‐analysis . Diabetol Metab Syndr . 2024 ;16 :101 . doi:10.1186/s13098-024-01346-4
PMC11092065 38745237

7Ott A , Stolk RP , van Harskamp F , Pols HAP , Hofman A , Breteler MMB . Diabetes mellitus and the risk of dementia: the Rotterdam Study . Neurology . 1999 ;53 :1937 ‐1937 . doi:10.1212/WNL.53.9.1937
10599761

8Arvanitakis Z , Wilson RS , Bienias JL , Evans DA , Bennett DA . Diabetes Mellitus and Risk of Alzheimer Disease and Decline in Cognitive Function . Arch Neurol . 2004 ;61 :661 ‐666 . doi:10.1001/archneur.61.5.661
15148141

9Ding X , Yin Li , Zhang L , et al. Diabetes accelerates Alzheimer's disease progression in the first year post mild cognitive impairment diagnosis . Alzheimer's & Dementia . 2024 ;20 :4583 ‐4593 . doi:10.1002/alz.13882
PMC11247667 38865281

10Domínguez RO , Pagano MA , Marschoff ER , González SE , Repetto MG , Serra JA . Alzheimer disease and cognitive impairment associated with diabetes mellitus type 2: associations and a hypothesis . Neurología (English Edition) . 2014 ;29 :567 ‐572 . doi:10.1016/j.nrleng.2014.10.001
24140159

11Sanz C , Andrieu S , Sinclair A , Hanaire H , Vellas B . Diabetes is associated with a slower rate of cognitive decline in Alzheimer disease . Neurology . 2009 ;73 :1359 ‐1366 . doi:10.1212/WNL.0b013e3181bd80e9
19858457

12Bloom GS . Amyloid‐β and Tau: the Trigger and Bullet in Alzheimer Disease Pathogenesis . JAMA Neurol . 2014 ;71 :505 ‐508 . doi:10.1001/jamaneurol.2013.5847
PMC12908160 24493463

13DeTure MA , Dickson DW . The neuropathological diagnosis of Alzheimer's disease . Mol Neurodegener . 2019 ;14 :32 . doi:10.1186/s13024-019-0333-5 PMC6679484 31375134

14van der Kant R , Goldstein LSB , Ossenkoppele R . Amyloid‐β‐independent regulators of tau pathology in Alzheimer disease . Nat Rev Neurosci . 2020 ;21 :21 ‐35 .10.1038/s41583-019-0240-3 31780819

15van Arendonk J , Neitzel J , Steketee RME , et al. Diabetes and hypertension are related to amyloid‐beta burden in the population‐based Rotterdam Study . Brain . 2023 ;146 :337 ‐348 .10.1093/brain/awac354 PMC9825526 36374264

16Moran C , Beare R , Phan TG , et al. Type 2 diabetes mellitus and biomarkers of neurodegeneration . Neurology . 2015 ;85 :1123 ‐1130 . doi:10.1212/WNL.0000000000001982
PMC5573049 26333802

17Roberts RO , Knopman DS , Cha RH , et al. Diabetes and Elevated Hemoglobin A1c Levels Are Associated with Brain Hypometabolism but Not Amyloid Accumulation . J Nucl Med . 2014 ;55 :759 ‐764 . doi:10.2967/jnumed.113.132647
PMC4011952 24652830

18van Gils V , Jansen WJ , van der Flier WM , et al. The association of diabetes with Alzheimer's disease biomarkers and vascular burden across European aging and memory clinic cohorts . Alzheimer's & Dementia . 2025 ;21 :e70804.10.1002/alz.70804 PMC12573102 41165072

19Lan G , Wang M , Gonzalez‐Ortiz F , et al. Comprehensive evaluation of plasma tau biomarkers for detecting and monitoring Alzheimer's disease in a multicenter and multiethnic aging population . Nat Aging . 2025 ;5 :1601 ‐1618 . doi:10.1038/s43587-025-00904-3
40550877

20Palmqvist S , Janelidze S , Quiroz YT , et al. Discriminative Accuracy of Plasma Phospho‐tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders . JAMA . 2020 ;324 :772 ‐781 . doi:10.1001/jama.2020.12134
PMC7388060 32722745

21Mattsson‐Carlgren N , Salvadó G , Ashton NJ , et al. Prediction of Longitudinal Cognitive Decline in Preclinical Alzheimer Disease Using Plasma Biomarkers . JAMA Neurol . 2023 ;80 :360 ‐369 . doi:10.1001/jamaneurol.2022.5272
PMC10087054 36745413

22Salvadó G , Janelidze S , Bali D , et al. plasma phosphorylated tau 217 to identify preclinical Alzheimer disease . JAMA Neurol . 2025 ;82 :1122 ‐1134 . doi:10.1001/jamaneurol.2025.3217
PMC12558403 40952756

23Cui L , Huang L , Pan F‐F , et al. Chinese Preclinical Alzheimer's Disease Study (C‐PAS): design and challenge from PET Acceptance . J Prev Alzheimer's Dis . 2023 ;10 :571 ‐580 . doi:10.14283/jpad.2023.49
37357299

24Wang J , He Y , Chen X , et al. Metabotropic glutamate receptor 5 (mGluR5) is associated with neurodegeneration and amyloid deposition in Alzheimer's disease: a [18F]PSS232 PET/MRI study . Alz Res Therapy . 2024 ;16 :9 . doi:10.1186/s13195-024-01385-z
PMC10785459 38217040

25Luan Y , Wang W , Huang Qi , et al. Synaptic loss pattern is constrained by brain connectome and modulated by phosphorylated tau in Alzheimer's disease . Nat Commun . 2025 ;16 :6356 . doi:10.1038/s41467-025-61497-4
PMC12246497 40640137

26Bondi MW , Edmonds EC , Jak AJ , et al. Neuropsychological Criteria for Mild Cognitive Impairment Improves Diagnostic Precision, Biomarker Associations, and Progression Rates . J Alzheimer's Dis . 2014 ;42 :275 ‐289 . doi:10.3233/JAD-140276
PMC4133291 24844687

27Mao X , Wang Y , Luan Y , et al. Interplay between astrocyte reactivity and APOE ε4 status is associated with accelerated pTau‐related tau pathology in Alzheimer's disease . Mol Neurodegener . 2025 ;20 :115 . doi:10.1186/s13024-025-00906-0
PMC12573940 41163175

28Petersen RC , Aisen PS , Beckett LA , et al. Alzheimer's Disease Neuroimaging Initiative (ADNI) . Neurology . 2010 ;74 :201 ‐209 . doi:10.1212/WNL.0b013e3181cb3e25
PMC2809036 20042704

29O'Bryant SE , Johnson LA , Barber RC , et al. The Health & Aging Brain among Latino Elders (HABLE) study methods and participant characteristics . Alzheimers Dement (Amst) . 2021 ;13 :e12202. doi:10.1002/dad2.12202
PMC8215806 34189247

30Brum WS , Ashton NJ , Simrén J , et al. Biological variation estimates of Alzheimer's disease plasma biomarkers in healthy individuals . Alzheimer's & Dementia . 2024 ;20 :1284 ‐1297 . doi:10.1002/alz.13518
PMC10916965 37985230

31Abner EL , Nelson PT , Kryscio RJ , et al. Diabetes is associated with cerebrovascular but not Alzheimer's disease neuropathology . Alzheimer's & Dementia . 2016 ;12 :882 ‐889 . doi:10.1016/j.jalz.2015.12.006
PMC4958598 26812281

32Ahtiluoto S , Polvikoski T , Peltonen M , et al. Diabetes, Alzheimer disease, and vascular dementia . Neurology . 2010 ;75 :1195 ‐1202 . doi:10.1212/WNL.0b013e3181f4d7f8
20739645

33Arvanitakis Z , Schneider JA , Wilson RS , et al. Diabetes is related to cerebral infarction but not to AD pathology in older persons . Neurology . 2006 ;67 :1960 ‐1965 . doi:10.1212/01.wnl.0000247053.45483.4e
17159101

34dos Santos Matioli MNP , Suemoto CK , Rodriguez RD , et al. Diabetes is Not Associated with Alzheimer's Disease Neuropathology . J Alzheimers Dis . 2017 ;60 :1035 ‐1043 . doi:10.3233/JAD-170179
PMC5892204 28984587

35Sadrolashrafi K , Craft S , Decourt B , et al. Is diabetes associated with increased pathological burden in Alzheimer's disease? . Alzheimers Dement (Amst) . 2021 ;13 :e12248. doi:10.1002/dad2.12248
PMC8579894 34796262

36Chiang GC , Chang E , Pandya S , et al. Cognitive deficits in non‐demented diabetic elderly appear independent of brain amyloidosis . J Neurol Sci . 2017 ;372 :85 ‐91 . doi:10.1016/j.jns.2016.11.040
PMC5193167 28017255

37Santiago JA , Karthikeyan M , Lackey M , Villavicencio D , Potashkin JA . Diabetes: a tipping point in neurodegenerative diseases . Trends Mol Med . 2023 ;29 :1029 ‐1044 . doi:10.1016/j.molmed.2023.09.005
PMC10844978 37827904

38Vermeer SE , Sandee W , Algra A , Koudstaal PJ , Kappelle LJ , Dippel DWJ , Impaired Glucose Tolerance Increases Stroke Risk in Nondiabetic Patients With Transient Ischemic Attack or Minor Ischemic Stroke . Stroke 37 , 1413 ‐1417 (2006 ). doi:10.1161/01.STR.0000221766.73692.0b
16627787

39Barini E , Antico O , Zhao Y , et al. Metformin promotes tau aggregation and exacerbates abnormal behavior in a mouse model of tauopathy . Mol Neurodegener . 2016 ;11 :16 . doi:10.1186/s13024-016-0082-7
PMC4746897 26858121

40Kickstein E , Krauss S , Thornhill P , et al. Biguanide metformin acts on tau phosphorylation via mTOR/protein phosphatase 2A (PP2A) signaling . Proc Natl Acad Sci . 2010 ;107 :21830 ‐21835 . doi:10.1073/pnas.0912793107
PMC3003072 21098287

41Chen Y , Zhao S , Fan Z , et al. Metformin attenuates plaque‐associated tau pathology and reduces amyloid‐β burden in APP/PS1 mice . Alz Res Therapy . 2021 ;13 :40 . doi:10.1186/s13195-020-00761-9
PMC7871393 33563332

42Ou Z , Kong X , Sun X , et al. Metformin treatment prevents amyloid plaque deposition and memory impairment in APP/PS1 mice . Brain, Behavior, and Immunity . 2018 ;69 :351 ‐363 . doi:10.1016/j.bbi.2017.12.009
29253574

43Chen Y , Zhou K , Wang R , et al. Antidiabetic drug metformin (GlucophageR) increases biogenesis of Alzheimer's amyloid peptides via up‐regulating BACE1 transcription . Proc Natl Acad Sci . 2009 ;106 :3907 ‐3912 . doi:10.1073/pnas.0807991106
PMC2656178 19237574

44Tahmi M , Luchsinger JA . Metformin in the Prevention of Alzheimer's Disease and Alzheimer's Disease Related Dementias . J Prev Alzheimer's Dis . 2023 ;10 :706 ‐717 . doi:10.14283/jpad.2023.113
37874091

45Sabbagh MN , et al. Repurposing glucagon‐like peptide‐1 receptor agonists for the treatment of neurodegenerative disorders . Nat Aging . 2025 ;6 :56 –67 . doi:10.1038/s43587-025-01029-3
PMC13528608 41419667

46Seminer A , Mulihano A , O'Brien C , et al. Cardioprotective Glucose‐Lowering Agents and Dementia Risk . JAMA Neurol . 2025 ;82 :450 ‐460 . doi:10.1001/jamaneurol.2025.0360
PMC11976645 40193122

47Zhang Y , Chen H , Feng Y , et al. Activation of AMPK by GLP‐1R agonists mitigates Alzheimer‐related phenotypes in transgenic mice . Nat Aging . 2025 ;5 :1097 ‐1113 . doi:10.1038/s43587-025-00869-3
40394225

48McClean PL , Hölscher C . Liraglutide can reverse memory impairment, synaptic loss and reduce plaque load in aged APP/PS1 mice, a model of Alzheimer's disease . Neuropharmacology . 2014 ;76 :57 ‐67 . doi:10.1016/j.neuropharm.2013.08.005
23973293

49McClean PL , Parthsarathy V , Faivre E , Hölscher C . The Diabetes Drug Liraglutide Prevents Degenerative Processes in a Mouse Model of Alzheimer's Disease . J Neurosci . 2011 ;31 :6587 ‐6594 . doi:10.1523/JNEUROSCI.0529-11.2011
PMC6622662 21525299

50Guo Z , Chen Y , Mao Y‐F , et al. Long‐term treatment with intranasal insulin ameliorates cognitive impairment, tau hyperphosphorylation, and microglial activation in a streptozotocin‐induced Alzheimer's rat model . Sci Rep . 2017 ;7 :45971. doi:10.1038/srep45971
PMC5382700 28382978

51Bu X‐Le , Zhu W , Wang Q‐H , et al. Association between self‐reported multimorbidity and longitudinal brain Aβ deposition in Alzheimer's disease . Nat Commun . 2025 ;16 :5905 . doi:10.1038/s41467-025-60748-8
PMC12219109 40592807

Come citare questo studio

Min Wu, Ying Luan, Xiaowen Xu, Xiaoxie Mao, Huiying Lin, Kun He, Qi Huang, Jie Zhang, Sheng Kuang, Yihui Guan, Alzheimer's Disease Neuroimaging Initiative, Health and Aging Brain Study (HABS‐HD) Team, Sid O'Bryant, Binyin Li, Qihao Guo, Fang Xie, Benyan Luo. (2026). Diabetes mellitus is associated with altered Alzheimer's disease pathology.. Alzheimer's & dementia : the journal of the Alzheimer's Association. DOI: 10.1002/alz.72002

Consulta la pubblicazione originale ↗