Gut-brain axis crosstalk in patients with Alzheimer's disease and high body mass index: Gut dysbiosis,blood-brain barrier disruption,and neuroinflammation

Abstract

Background

Accumulating evidence indicates that body mass index (BMI) is related to Alzheimer's disease (AD).

Objective

This study aimed to investigate the impact of high BMI on cognitive function in AD patients and potential mechanisms involving gut dysbiosis, blood-brain barrier (BBB) disruption, and neuroinflammation in brain.

Methods

A total of 50 AD patients were recruited and divided into AD with normal BMI (AD-nBMI) and AD with high BMI (AD-hBMI) groups. Cognitive function was assessed, the levels of BBB variables, neuroinflammatory factors, and AD biomarkers in cerebrospinal fluid (CSF) were measured, gut microbiota and metabolites were analyzed using 16S ribosomal ribonucleic acid gene sequencing and gas chromatography-mass spectrometry analysis, and the correlations among gut microbiota and metabolites, BBB variables, and neuroinflammatory factors in CSF were analyzed.

Results

AD-hBMI group exhibited impaired overall cognition, elevated CSF levels of zonula occludens-1 (ZO-1) and occludin (OCLN) (BBB disruption), and increased CSF levels of nitric oxide (NO) and hydroxyl radical (·OH) (neuroinflammation) compared with AD-nBMI group. Correlation analyses revealed that BMI was positively associated with impaired cognition, CSF levels of OCLN and NO in AD patients. AD-hBMI group displayed a unique gut dysbiosis pattern characterized by the alterations in specific metabolite levels. In AD-hBMI group, elevated thioridazine, 4-hydroxybenzaldehyde, and N-acetylleucine were positively correlated with ZO-1 level in CSF, reduced acesulfame and undecanoic acid were negatively correlated with OCLN level in CSF, and elevated (S)-S-methylcysteine sulfoxide was positiveiy correlated with both ZO-1 and OCLN levels in CSF. Elevated thioridazine and medroxyprogesterone were positively correlated with NO level in CSF, and decreased butyric acid was negatively correlated with ·OH level in CSF.

Conclusions

High BMI may accelerate gut dysbiosis, leading to disrupted BBB, enhanced neuroinflammation, and ultimately accelerated cognitive impairment in AD patients.

Keywords

Alzheimer's disease blood-brain barrier body mass index cognitive impairment gut microbiota and metabolites neuroinflammation

Introduction

Alzheimer's disease (AD) is the most common neurodegenerative disorder characterized by progressively deteriorated cognitive function, exacerbated neuropsychiatric symptoms, and impaired activities of daily living. It accounts for 60%-80% of dementia cases and affects 11.3% of individuals aged 65 years and older.¹ The hallmarks of AD include the early neuroinflammation and subsequent extracellular deposition of amyloid plaques composed of amyloid-β (Aβ), followed by the intracellular development of neurofibrillary tangles composed of phosphorylated tau (P-tau).² AD is increasingly recognized as a complex illness caused by genetic predisposition, environmental factors, and unhealthy lifestyle.³ Therefore, early identification and intervention of preventable risk factors are very crucial for delaying the pathology and progression of AD.

It was reported that elevated body mass index (BMI) was associated with the increased risk of all-cause dementia, and obesity was associated with the increased risk of AD.⁴ Accumulating evidence identified obesity as a significant risk factor for the development of AD.^5–6 A rat model study found that long-term high-calorie, high-sugar, and low-fiber diets were among the leading causes of obesity, which accelerated cognitive impairment through chronic peripheral inflammation.⁷ AD model mice fed with high-fat diet yielded obesity and elevated the levels of Aβ and tau pathology in brain.⁸ A longitudinal study recruiting 6583 individuals showed that participants with larger sagittal abdominal diameter had a roughly threefold higher risk of dementia compared to those with smaller one.⁹ A cross-sectional analysis of an epidemiologic cohort study of cognitive and functional decline revealed that the increased waist-hip ratio was linked to the decreased hippocampal volume.¹⁰ Nevertheless, the specific mechanisms by which obesity accelerates cognitive impairment in AD patients have not yet been elucidated.

Disrupted blood-brain barrier (BBB) may serve as a critical link between obesity and AD. Obesity was observed to impair cognitive function by disrupting BBB^11–12 and establish the cross-talk between peripheral and central neuroinflammation.¹³ BBB consists of microvascular endothelial cells, end-feet of astrocytes, and pericytes. Microvascular endothelial cells are glued together by tight-junction proteins and scaffolding proteins.¹⁴ When BBB structure is intact, tight junction proteins are stably expressed in the tight junctions of cerebrovascular endothelial cells. When BBB is disrupted due to pathological conditions, tight junction was broken down, causing these proteins to detach from junction and be released into brain interstitial fluid, leading to their elevated levels in cerebrospinal fluid (CSF).^15–16 In mice fed with high-fat diet, BBB integrity disruption, marked by extravasated plasma-derived IgG, was associated with intensified neuroinflammation and hippocampus-dependent cognitive decline.¹⁷ The disrupted BBB integrity induced by high-fat diet was counteracted by lipid-lowering and anti-inflammatory agents in mice.¹⁸ Experiments in AD animal models showed that the disruption of BBB tight junction exacerbated neuroinflammation in brain, leading to a cascade of neuronal damage and cognitive decline.¹⁹ Nevertheless, the role of gut microbiota dysbiosis in obesity-induced BBB disruption and subsequent neuroinflammation remains unclear.

Gut-brain axis dysfunction influenced the cognitive function of obese subjects.²⁰ Gut microbiota dysbiosis was implicated as a peripheral inflammatory indicator of obesity-accelerated cognitive impairment due to its unique susceptibility to lifestyle and environmental factors.^21–23 The metabolites of gut microbiota, such as free fatty acids, were found to cause peripheral inflammation, which played an important role in triggering neuroinflammation featured by the glial over-activation in brain.²⁴ A pivotal study demonstrated that transplanting gut microbiota from AD patients into germ-free AD mice model resulted in enrichment of bacteroides and increased polyunsaturated fatty acid through multi-omics analyses, which activated microglia-mediated neuroinflammation, ultimately exacerbating AD pathological deposition.²⁵ However, the species differences in gut microbiota and BBB structure between mice and humans restrict the direct extrapolation of the findings to clinical applications. Therefore, it remains unclear how to explore the potential mechanisms of AD with obesity involving gut microbiota dysbiosis, BBB disruption, and neuroinflammation in the brain of human population.

In this study, a total of 50 AD patients were recruited and divided into AD with normal BMI (AD-nBMI) and AD with high BMI (AD-hBMI) groups. Demographic data were collected and cognitive function was assessed by standardized rating scales, the levels of BBB variables, neuroinflammatory factors, and AD biomarkers in CSF were measured by enzyme-linked immunosorbent assay (ELISA) or chemical colorimetry, gut microbiota and metabolites in fecal samples were analyzed using 16S rRNA gene sequencing and gas chromatography-mass spectrometry (GC-MS) analyses, correlations among the above variables were analyzed, and the potential effect of disturbed gut metabolites in mediating the association between BBB disruption and neuroinflammation in brain was finally analyzed. This study aimed to elucidate the mechanistic pathway from high BMI to AD, hypothesizing that gut microbiota dysbiosis might serve as a critical initiator, leading to BBB disruption and ultimately driving neuroinflammation in brain.

Methods

Ethics statement

This investigation received ethical approval from the Institutional Review Board of Beijing Tiantan Hospital, Capital Medical University (KY2024-260-03). In line with the Declaration of Helsinki, written informed consent forms were duly signed by all participants and their family members.

Inclusion and exclusion criteria for AD patients

AD patients at different stages were enrolled from the Cohort study of AD patients in Beijing Tiantan Hospital, Capital Medical University (Supplemental Figure 1). AD patients were diagnosed according to the National Institute on aging-Alzheimer's Association 2018 new research framework (2018-NIA-AA-RF).²⁶

The exclusion criteria for this study were as follows: (1) History of stroke temporally associated with the onset or exacerbation of cognitive impairment, or the presence of multiple or extensive infarcts, or severe white matter hyperintensity. (2) Cognitive impairments attributable to neurological conditions other than AD, such as frontotemporal degeneration, Lewy body disease, corticobasal degeneration, Parkinson's disease, multiple sclerosis, and epilepsy, etc. (3) The presence of diseases that might significantly affect cognitive function, such as non-neurological diseases or a history of drug use. (4) Articulation disorders, depression, and mental illnesses affecting emotional expression. (5) Inflammatory and infectious diseases or use of antibiotics. (6) The presence of hearing loss, visual impairment, or illiteracy that might affect completion of assessments.

Groups

BMI was calculated as body weight in kilograms divided by squared height in meters (kg/m²). Following the recommendations of the Working Group on Obesity in China for Chinese population,²⁷ the recruited AD patients were divided into AD with normal BMI (AD-nBMI, BMI 18.5-23.9 kg/m²) group and AD with high BMI (AD-hBMI, BMI≥24 kg/m²) group according to their levels of BMI.

Collection of demographic information

Demographic data, encompassing age, sex, educational level, apolipoprotein E (APOE) ε4 carrier status, smoking, drinking, constipation, BMI, water intake, and history of hypertension, diabetes, and coronary heart disease, etc., were recorded from AD patients.

Assessments of cognitive function

Overall cognitive function was assessed by Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scales. MMSE scale assesses various cognitive domains, including orientation, recall, attention, computation, memory, and language. Patients with illiteracy, primary education, or more education than junior college have cognitive impairment when the scale score is less than 17, 20, or 24 points, respectively.²⁸ MMSE scale is sensitive to dementia. The lower the score of MMSE scale, the more severe the overall cognitive impairment. MoCA scale evaluates multiple cognitive domains, comprising visuospatial ability, executive function, naming, memory, attention, language, delayed recall and orientation. Cognitive impairment is indicated as a MoCA scale score of 26 points or lower. Individuals with less than 12 years of education receive additional 1 point.²⁹ MoCA scale is sensitive to mild cognitive impairment. The lower the score of MoCA scale, the more severe the overall cognitive impairment.

Measurements of the levels of BBB variables, neuroinflammatory factors, and AD biomarkers in CSF

Anti-AD drugs were withdrawn for 12–14 h if patient's condition allowed, and a long time was considered unethical by our ethics committee. Under fasting conditions, 5 ml of CSF was collected in a polypropylene tube throughlumbar puncture between 7 and 9 a.m. CSF samples were centrifuged immediately at 3000 rpm, aliquoted into separate Nunc cryotubes, and frozen at −80°C until assay.

In the neuroinflammatory factors detected in the CSF from AD-hBMI and AD-nBMI groups, the levels of interleukin-1β (IL-1β) (Human IL-1 beta Uncoated ELISA kit, Invitrogen, USA), soluble triggering receptor expressed on myeloid cells 2 (sTREM2) (Human sTREM2 DuoSet ELISA, R&D Systems, USA), glial fibrillary acidic protein (GFAP) (NS830kit, Merck Millipore Company, Germany), and YKL-40 (also known as chitinase-3-like protein 1, CHI3L1) (ProcartaPlex™ Immunoassay Kit, Invitrogen, USA) were measured by enzyme-linked immunosorbent assay (ELISA); the levels of nitric oxide (NO) (A013 kit, Nanjing Jiancheng Biological Engineering Research Institute, China), hydrogen peroxide (H₂O₂) (A064 kit, Nanjing Jiancheng Biological Engineering Research Institute, China), and hydroxy radical (·OH) (A018 kit, Nanjing Jiancheng Biological Engineering Research Institute, China) were measured by chemical colorimetry. In the BBB variables detected in the CSF from AD-hBMI and AD-nBMI groups, the levels of zonula occludens-1 (ZO-1) (MBS2605490 kit, MyBioSource Company, USA), occludin (OCLN) (NBP2-80305 kit, Novus Biologicals Company, USA), claudin-5 (CLDN5) (NBP2-75332 kit, Novus Biologicals Company, USA), receptor for advanced glycation end-products (RAGE) (DRG00 kit, R&D systems Company, USA), and low-density lipoprotein receptor-related protein 1(LRP1) (MBS772326 kit, MyBioSource Company, USA) were measured by ELISA.

In the AD biomarkers detected in the CSF from AD-hBMI and AD-nBMI groups, the levels of Aβ42 (CSB-E10684 h kit, CUSABIO Company, China), P-tau181 (KHO0631 kit, Invitrogen Company, USA), P-tau199 (KHB8051 kit, Invitrogen Company, USA), P-tau231 (KHB8051 kit, Invitrogen Company, USA), P-tau396 (KHB7031 kit, Invitrogen Company, USA), and total tau (T-tau) (CSB-E12011 h kit, CUSABIO Company, China) were measured by ELISA.

The specific testing procedure for each of above variables followed the corresponding instruction provided by individual kit.

Fecal sample collection, deoxyribonucleic acid (DNA) extraction and 16S ribosomal ribonucleic acid (rRNA) gene sequencing

Fresh fecal samples from AD-nBMI and AD-hBMI groups were collected and preserved at −80°C for assay.

E.Z.N.A.^® Stool DNA Kit (OMEGA Bio-tek, Norcross, USA) was used to extract overall genomic DNA from these samples. DNA quality was tested using Qubit Fluorometer dsDNA BR Assay Kit (Invitrogen, Carlsbad, USA). Using particular primers, the V3-V4 sections of bacterial 16S rRNA genes were amplified for α and β diversity studies.

Phusion^® High-Fidelity polymerase chain reaction (PCR) Master Mix (New England Biolabs, Ipswich, USA) was used for all PCR reactions. Gel Extraction Kit (Qiagen, Hilden, Germany) was used to normalize and purify PCR products. The AmpliSeq for Illumina Library Prep Kit (Illumina, San Diego, USA) was used to prepare and index the library. On the Illumina HiSeq 2500 platform, HiSeq PE Cluster Kit v4 cBot and HiSeq SBS Kit v4 (Illumina, San Diego, USA) were utilized to generate 250 bp paired-end reads.

After sequencing, low-quality reads were filtered out to keep clean data for analysis. Tags were created from overlapped reads. Uparse v7.0.1001 (http://www.drive5.com/uparse/) grouped tags with 97% similarity into operational taxonomic units (OTUs). The usearch_global function assigned tags to typical OTUs sequences. Species annotation was done by matching OTUs representative sequences against a database using Ribosomal Database Project Classifier (v2.2). OTUs representative sequences were used to determine software confidence thresholds of 0.8 for tags longer than 250 bp and 0.5 for shorter tags. OTUs and species annotations were used to analyze sample species complexity and inter-group species differences.

Metabolomic analyses of fecal samples

After homogenizing and centrifuging 10 mg fecal sample, the supernatant was filtered and derivatized before analysis. Procedure was as follows: To a 10 mg fecal sample, 20 μL deionized water, 120 μL sample release agent, and two metabolic analysis-specific steel beads were added. The samples were homogenized in a tissue homogenizer at 50 Hz for 300 s and centrifuged at 4°C and 18,000 g for 20 min. After centrifugation, the fecal sample supernatant was filtered and 30 μL was put to a 96-well plate with 20 μL derivatization reagent and 20 μL 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide working solution. A 96-well plate was sealed with aluminum film and reacted for 60 min at 40°C and 1200 rpm. A reaction plate was centrifuged for 10 min at 4°C and 4000 g after reaction. After transferring 30 μL sample to a new 96-well plate, 90 μL pre-cooled sample diluent was added. The mixture was homogenized for 5 min at 600 rpm and 10°C. After 5 min of centrifugation at 4°C and 4000 g, the supernatants were ready for gas chromatography-mass spectrometry (GC-MS) analysis.

Metabolites of gut microbiota were separated and quantified using a Waters ACQUITY UPLC I-Class Plus system coupled with a QTRAP 6500 Plus high-sensitivity mass spectrometer (SCIEX, Framingham, USA). Chromatographic column employed was a BEH C18 (2.1 mm×10 cm, 1.7 µm, Waters, Milford, USA). Mobile phase consisted of 0.1% formic acid in water (Solvent A) and 30% isopropanol in acetonitrile (Solvent B). Mass spectrometric conditions for QTRAP 6500 Plus equipped with an ESI Turbo Ion-Spray interface were as follows: ion source temperature was set to 400°C; ion spray voltage (IS) was set at 4500 V for positive mode and −4500 V for negative mode; ion source gases I (GS1), II (GS2), and curtain gas (CUR) were set at 60, 60, and 35 pounds per square inch, respectively. Quantitative data and metabolite identification results were compiled into a data matrix using Skyline quantitative software, which was subsequently analyzed for information processing.

Clinical data analyses

Statistical analyses were performed using R software (version 4.3.1). p < 0.05 was considered statistically significant.

Descriptive results were expressed as means ± standard deviations (SDs) or medians with interquartile ranges (Q1-Q3), depending on the distribution characteristics of data. Student t test (2 groups) was used for the normally distributed variables with equal population variance, and non-parametric Mann-Whitney U (2 groups) test was used for the variables with non-normal distributions. Chi-squared test was used for the comparisons of categorical variables. For variables with significant differences, box plots were generated using ggplot2 package in R software to illustrate the between-group differences. Linear and hierarchical regressions were performed for the variables that showed a correlation with significantly different clinical variables and metabolites. Variables were adjusted for age, sex, education level, smoking, drinking, constipation, water intake, hypertension, diabetes, and coronary heart disease.

Spearman's rank correlation analysis was conducted to explore the correlations among BBB variables, neuroinflammatory factors, and AD biomarkers in the CSF from AD-hBMI and AD-nBMI groups. Significant differential features between the two groups were visualized through heatmaps using R package Complex Heatmap.

Gut microbiota data analyses

16S rRNA gene sequencing data were analyzed through nf-core/ampliseq v2.7.0 pipeline, which integrated all necessary analysis steps and software, and was publicly accessible at https://nf-co.re/ampliseq. All data analyses were primarily conducted in R 4.3.1 environment using phyloseq package (version 1.44.0) and microbiome package (version 1.22.0).

α diversity refers to the richness and diversity within a single sample and is calculated based on OTUs table and phylogenetic tree. In the analysis of α diversity, the amplicon sequence variants in less than 10% of samples were removed. Diversity indices, such as Chao1 index, Observed index, Simpson index, and Shannon index, were calculated using filtered phyloseq objects. In the two group differential abundance analysis, centered log-ratio transformation or total sum scaling method was applied.

β diversity analysis was used to calculate differences in microbial community composition between samples, which were visualized through non-metric multidimensional scaling (NMDS). Statistically significant difference was validated with permutational multivariate analysis of variance (PERMANOVA) testing.

Taxonomic composition analysis was performed using the function transform_taxonomy from R package microbiota, calculating the relative abundance of each taxonomic unit at different levels of classification, and showing the top 20 most abundant taxonomic units. A linear discriminant analysis effect size (LEfSe) was used to identify the microbial taxonomic units with significant differences between AD-hBMI and AD-nBMI groups, setting the LDA threshold greater than 2 and p value less than 0.05. The LEfSe analysis employs the non-parametric Kruskal-Wallis test, which is more sensitive for detecting differences in specific taxa between groups. It is particularly suitable for research scenarios where overall community differences are not significant, but local variations exist.

Differences in microbial interactions were shown through the correlation network analysis of gut microbiota community, visualized with the network graph.

Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses and MetaCyc metabolic pathway database analyses were conducted, revealing the potential biological pathways relating to the pathogenesis of AD in which these gut microbiota might be involved.

Gut metabolomics data analyses

Partial least squares discriminant analysis (PLS-DA) was used to identify the key metabolites that distinguished AD-hBMI group from AD-nBMI group. The top 15 metabolites were ranked by random forest model based on their relative abundances and significance in differentiating AD-hBMI group from AD-nBMI group.

Multi-omics data integration analyses

Spearman correlation coefficients and permutation tests were calculated to assess the statistical significance of the correlations between gut microbiota and metabolites.

Results

A total of 100 participants were initially screened for eligibility, with 23 excluded for failing to meet the inclusion criteria. Of the remaining participants, 77 were eligible and consented to take part in the study. During the course of the study, 23 participants were excluded due to incomplete CSF data, and 4 were excluded because of their unqualified fecal samples (Supplemental Figure 1). In the end, a total of 50 AD patients were enrolled in this study.

Comparisons of demographic information between AD-hBMI and AD-nBMI groups

Comparisons of demographic data showed that the average BMI was 21.39 kg/m² in AD-nBMI group and 25.82 kg/m² in AD-hBMI group. There were no significant differences between the two groups in terms of age, sex, education level, APOE ε4 carrier, smoking, drinking, constipation, water intake more than 1L, and history of hypertension, diabetes, and coronary heart disease (Table 1).

Table 1.

Demographic information and cognitive function of AD-hBMI and AD-nBMI groups.

	Total AD patients (n = 50)	AD-nBMI group (n = 30)	AD-hBMI group (n = 20)	p
Demographic information
Age (years, mean ± SD)	60.32 ± 11.03	58.5 ± 11.49	61.85 ± 7.70	0.259
Female [n (%)]	30 (60%)	20 (66.7%)	10 (50%)	0.239
Low educational level (less than 9 years) [n (%)]	11 (22%)	5 (16.7%)	6 (30%)	0.265
APOE ε4 carrier [n (%)]	17 (50%)	9 (40.9)	8 (66.7%)	0.151
Smoking [n (%)]	17 (34%)	10 (33.3%)	7 (35%)	0.903
Drinking (n, %)	13 (26%)	6 (20%)	7 (35%)	0.236
Constipation [n (%)]	7 (14%)	4 (13.3%)	3 (15%)	0.868
BMI (kg/m², mean ± SD)	22.87 ± 2.86	21.39 ± 1.36	25.82 ± 1.59	0.000***
Water intake more than 1 liters [n (%)]	21 (44.0%)	16 (53.3%)	6 (30.0%)	0.103
Hypertension [n (%)]	21 (42.0%)	11 (36.7%)	10 (50.0%)	0.349
Diabetes [n (%)]	5 (10.0%)	3 (10.0%)	2 (10.0%)	1.000
Coronary heart disease [n (%)]	8 (16.0%)	6 (20.0%)	2 (10.0%)	0.345
Cognitive function
MMSE (points, mean ± SD)	14.92 ± 5.95	16.48 ± 5.15	12.65 ± 6.43	0.025*
MoCA (points, mean ± SD)	11.24 ± 5.74	12.69 ± 5.81	9.15 ± 5.10	0.033*

*p < 0.05, ***p < 0.001. Data were presented as number (percentage), means ± SD or median (quartile). AD-nBMI: Alzheimer's disease with normal body mass index (BMI 18.5-23.9 kg/m²); AD-hBMI: Alzheimer's disease with high body mass index (BMI≥24 kg/m²); AD: Alzheimer's disease; SDs: standard deviations; APOE: apolipoprotein E; BMI: body mass index; MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Comparisons of overall cognitive function between AD-hBMI and AD-nBMI groups

Comparisons of overall cognitive function indicated that AD-hBMI group had significantly lower scores of MMSE and MoCA scales compared with AD-nBMI group (both p < 0.05), indicating the significantly compromised overall cognitive function in AD-hBMI group (Figure 1A, B). Further, the correlations of BMI with the scores of overall cognition rating scales in AD patients were analyzed by linear regression. After adjusting for age, sex, educational level, smoking, drinking, constipation, water intake, history of hypertension, diabetes, and coronary heart disease, BMI was negatively correlated with the scores of MMSE and MoCA scales (all p < 0.05), indicating that AD patients with higher BMI had a worse overall cognitive function (Supplemental Table 1).

Comparisons of BBB variables and neuroinflammatory factors in CSF between AD-hBMI and AD-nBMI groups

The levels of BBB variables in CSF were compared between AD-hBMI and AD-nBMI groups. It was found that AD-hBMI group had significantly increased levels of ZO-1 and OCLN in CSF compared with AD-nBMI group (both p < 0.05) (Supplemental Table 2 and Figure 1C, D), indicating more severe BBB disruption in AD-hBMI group. Further, the correlations of BMI with the levels of neuroinflammatory factors in CSF that differed between the two groups were performed by linear regression. After adjusting for age, sex, educational level, smoking, drinking, constipation, water intake, history of hypertension, diabetes, and coronary heart disease, BMI was positively correlated with OCLN level in CSF (p < 0.05), indicating that AD patients with higher BMI had a more pronounced BBB disruption (Supplemental Table 3).

The levels of neuroinflammatory factors in CSF were compared between AD-hBMI and AD-nBMI groups. It was observed that sTREM2 level was significantly decreased and the levels of NO and ·OH were significantly increased in AD-hBMI group compared with AD-nBMI group (all p < 0.05) (Supplemental Table 2 and Figure 1E-G). Further, the correlations of BMI with the levels of neuroinflammatory factors in CSF that differed between the two groups were performed by linear regression. After adjusting for age, sex, educational level, smoking, drinking, constipation, water intake, history of hypertension, diabetes, and coronary heart disease, BMI was positively correlated with NO level and negatively correlated with sTREM2 level in CSF (all p < 0.05), demonstrating that AD patients with higher BMI had intensified neuroinflammation (Supplemental Table 3).

The correlations among the levels of BBB variables, neuroinflammatory factors, and AD biomarkers in the CSF from AD patients

The correlations of the levels of BBB variables with neuroinflammatory factors in CSF were assessed in AD patients. It was indicated that ZO-1 level was positively correlated with NO level in CSF. CLDN5 level was positively correlated with H₂O₂ level in CSF. RAGE level was positively correlated with H₂O₂ and IL-1β levels in CSF. GFAP level was positively correlated with H₂O₂ and IL-1β levels in CSF (all p < 0.05). The above results suggested that accelerated BBB disruption was associated with elevated levels of neuroinflammation in the brains of AD patients (Figure 2 and Supplemental Table 4).

The correlations between neuroinflammatory factors and AD biomarkers in CSF were conducted in AD patients. It was discovered that H₂O₂ level was negatively correlated with Aβ42 level in CSF, H₂O₂ level was positively correlated with P-tau231 and T-tau levels in CSF, and YKL-40 level was positively correlated with P-tau181 level in CSF (all p < 0.05). The above results implied that intensified neuroinflammation was associated with elevated levels of AD biomarkers in the brains of AD patients (Figure 2 and Supplemental Table 4).

The changes in gut microbiota of AD-hBMI and AD-nBMI groups

BMI did not significantly impact α diversity and β diversity in AD patients

α diversity index reflects the quantity and relative abundance of microbial communities. Within α diversity indices, Chao1 index and Observed index represent community richness, indicating the degree of richness in microbial community. Shannon index and Simpson index represent community diversity, reflecting comprehensive status of richness and evenness. In this study, compared with AD-nBMI group, AD-hBMI group showed no significant differences in Chao1 index, Observed index, Shannon index, and Simpson index (all p > 0.05), which indicated no significant differences in richness and evenness between the two groups (Figure 3A-D).

β diversity analysis was used to compare the gut microbiota community composition between the two groups. Based on NMDS, a greater distance between samples indicates a larger difference in community structure. The results showed no statistically significant difference in β diversity between AD-hBMI and AD-nBMI groups, suggesting no significant overall difference in the structure of gut microbiota between the two groups [PERMANOVA F = 0.795, p = 0.735] (Figure 3E).

High BMI altered the gut microbiota composition in AD patients

To explore the differences in gut microbiota composition between AD-hBMI and AD-nBMI groups, the composition and abundance of the top 20 microbiota in the community at genus level were analyzed. Their relative abundances were presented as percentages in stacked bar charts.

Compared with AD-nBMI group, the relative abundance of microbiota that was significantly decreased in AD-hBMI group included g_Gemmiger, g_Faecalibacterium, g_Streptococcus, g_Mediterraneibacter, g_Blautia, g_Roseburia, g_Bifidobacterium, g_Lachnospira, and g_Phocaeicola; the relative abundance of microbiota that was significantly increased in AD-hBMI group included g_Segatella, g_Akkermansia, g_Klebsiella, g_Escherichia, g_Romboutsia, g_Collinsella, g_Vescimonas, g_Anaerobutyricum, g_Ruminococcoides, and g_Bacteroides. These results indicated that high BMI altered the gut microbiota composition in AD patients (Figure 4A, B).

High BMI altered the gut microbiota characteristics in AD patients

To further explore the significant differences in gut microbiota in AD-hBMI group, LEfSe analysis was conducted. Compared to AD-nBMI group, AD-hBMI group had significantly increased relative abundances of g_Acidaminococcus, s_Acidaminococcus_fermentans, and s_Hoylesella_nanceiensis. In contrast, AD-hBMI group had significantly decreased relative abundances of p_Bacillota, s_Faecalibacillus_intestinalis, g_Faecalibacillus, g_Corynebacterium, f_Corynebacteriaceae, o_Mycobacteriales, g_Flintibacter, s_Flintibacter_butyricus, and s_Mediterraneibacter_butyricigenes. These results indicated that AD-hBMI group exhibited remarkably distinct gut microbiota characteristics (Figure 4C).

High BMI modified the interactions among gut microbiota in AD patients

To elucidate the impact of high BMI on gut microbiota in AD patients, a network analysis was conducted with aims to reveal structural changes within microbiota community and interpret the nature of microbial interactions. Red line indicated positive correlation, while green line indicated negative correlation. AD-nBMI group displayed a modular network distribution, lacking opportunistic pathogens, with a stable and orderly microbial community structure (Figure 5A). AD-hBMI group exhibited an over-connected and disordered network structure, with enrichment of opportunistic pathogens, such as g_Escherichia and g_Klebsiella, forming positively correlated chain-like structures (Figure 5B). These results indicated that BMI might be an important factor influencing the gut microbiota structure in AD patients.

The predictive functional profiling of microbiota communities of AD-hBMI and AD-nBMI groups

16S rRNA gene sequences were analyzed by using phylogenetic investigation of communities by reconstruction of unobserved states 2 (PICRUSt2), which allowed us to predict the significant differences in microbial metabolic functions between AD-hBMI and AD-nBMI groups.

In functional KEGG pathways, compared to AD-nBMI group, AD-hBMI group exhibited the notably increased activity of lipopolysaccharide biosynthesis, along with modestly increased activities of tropane, piperidine and pyridine alkaloid biosynthesis, isoquinoline alkaloid biosynthesis, and arginine and proline metabolism (Figure 5C). In MetaCyc pathway, AD-hBMI group displayed the enhanced metabolism of superpathway of pyrimidine deoxyribonucleosides degradation, L-phenylalanine biosynthesis, purine ribonucleosides degradation, O-antigen building blocks biosynthesis (E. coli), L-tyrosine biosynthesis, lipid IVA biosynthesis, purine deoxyribonucleosides degradation, and fatty acid elongation(saturated). AD-hBMI group showed the decreased metabolism of incomplete reductive TCA cycle, galactose degradation I (leloir pathway), acetylene degradation, Kdo transfer to lipid IVA III (chlamydia), pyrimidine ribonucleosides degradation, pyruvate fermentation to isobutanol (engineered), and glycogen biosynthesis I (from ADP-D-glucose) (Figure 5D). The different enriched pathways between the two groups suggested the potential metabolites that could mediate the impact of high BMI on AD patients through gut microbiota.

High BMI altered the composition of gut metabolites in AD patients

In this study, ultra-high performance liquid chromatography-mass spectrometry (UPLC-MS) was used to detect the changes in metabolites of AD-hBMI and AD-nBMI groups. The top 15 metabolites with significant changes were analyzed using PLS-DA. The results indicated a clear separation of metabolites between AD-nBMI and AD-hBMI groups, suggesting the significant differences in their metabolic profiles. The random forest model results showed the evidently elevated metabolites in AD-hBMI group, including hydroretrocortine, (S)C(S)S-S-methylcysteine sulfoxide, medroxyprogesterone, 2,3-diphosphoglyceric acid, 4-hydroxycinnamic acid, N-methyl-nicotinamide, ketoleucine, O-phosphorylethanolamine, and thioridazine. The obviously decreased metabolites in AD-hBMI group included 3,7-diketocholanic acid methyl ester, 3,4-dihydro-2h-1-benzopyran-2-one, 2-acetylpyrazine, benzoic acid, estriol, and acesulfame. These results demonstrated that high BMI altered the composition of gut metabolites in AD patients (Figure 6A, B).

The correlation between differential gut microbiota and differential metabolites in AD-nBMI and AD-hBMI groups

The correlation network diagram displayed the highly correlated gut microbiota and metabolites, with the top 13 microbial taxa between AD-hBMI and AD-nBMI groups selected based on their importance in LEfSe analysis and the top 30 metabolites selected based on random forest model between AD-hBMI and AD-nBMI groups. The results showed that N−methylnicotinamide level was positively correlated with the abundances of s_Faecalibacillus, s_Faecalibacillus_intestinalis, and o_Mycobacteriales; Butyric acid level was positively correlated with the abundances of s_Faecalibacillus, s_Faecalibacillus_intestinalis, p_Bacillota, and o_Mycobacteriales; Glucaric acid level was positively correlated with the abundances of o_Mycobacteriales, f_Corynebacteriaceae, and g_Corynebacterium; N−acetylleucine level was positively correlated with the abundances of p_Bacillota, o_Mycobacteriales, f_Corynebacteriaceae, and g_Corynebacterium; Dodecanedioic acid level was positively correlated with the abundances of s_Faecalibacillus, s_Faecalibacillus_intestinalis, and p_Bacillota; striol level was positively correlated with the abundances of s_Faecalibacillus and s_Faecalibacillus_intestinalis; Undecanoic acid level was negatively correlated with the abundances of s_Hoylesella_nanceiensis; Cis, cis−muconic acid level was negatively correlated with the abundances of s_Faecalibacillus and s_Faecalibacillus_intestinalis; The levels of (S)C(S)S−S−methylcysteine sulfoxide and N−acetylglutamine were negatively correlated with the abundance of s_Flintibacter_butyricus; O−phosphorylethanolamine level was negatively correlated with the abundances of f_Corynebacteriaceae and g_Corynebacterium; Itaconic acid level was negatively correlated with the abundance of o_Mycobacteriales. These findings elucidated the complex interactions between gut microbiota and metabolic profiles, providing insights into their mutual influence in AD patients with high BMI (Figure 7A and Supplemental Table 5).

Correlations of highly differential metabolites between AD-hBMI and AD-nBMI groups with BBB disruption and neuroinflammation in brain

We focused on the top 30 differential metabolites between AD-hBMI and AD-nBMI group identified by random forest model and analyzed their correlations with the levels of BBB variables and neuroinflammatory factors in CSF.

In AD-hBMI group, elevated thioridazine, 4-hydroxybenzaldehyde, and N-acetylleucine were positively correlated with ZO-1 level in CSF, reduced acesulfame and undecanoic acid were negatively correlated with OCLN level in CSF, and elevated (S)-S-methylcysteine sulfoxide was positiveiy correlated with both ZO-1 and OCLN levels in CSF.

Additionally, the metabolites that were positively correlated with the levels of NO (neuroinflammatory factor) in CSF included the increased levels of gut thioridazine and medroxyprogesterone in AD-hBMI group. The metabolites that were negatively correlated with the levels of ·OH (neuroinflammatory factor) in CSF included butyric acid in AD-hBMI group (Figure 7B and Supplemental Table 6).

Discussion

This study aimed to explore whether the disturbed gut microbiota and metabolites in AD-hBMI group promoted neuroinflammation in brain by disrupting BBB, worsening cognitive impairment. It was revealed that AD-hBMI patients exhibited significantly impaired cognitive function compared to AD-nBMI patients, and the severity of cognitive impairment was increased with BMI elevation. BBB was apparently disrupted, and evidently correlated with intensified neuroinflammation in AD-hBMI patients. High BMI induced dysbiosis of gut microbiota, and disturbed gut metabolites mediated the association between disrupted BBB and intensified neuroinflammation in brain, which eventually aggravate cognitive impairment in AD patients.

Cognitive impairment was associated with high BMI in AD patients

AD is the most common neurodegenerative disease among the elderly, andlacks effective interventions to slow its progression. Therefore, it is very imperative to identify and prevent risk factors to delay the pathology and progression of AD. High BMI is a critical risk factor for AD, which impairs the functions of multiple cognitive domains.³⁰ A cohort study involving 5693 middle and older adults from China Health and Retirement Longitudinal Study revealed significant differences in the trajectories of cognitive decline among individuals with different BMI levels, with obesity associated with a faster cognitive decline.³¹ A meta-analysis encompassing 72 studies with 4904 obese subjects revealed that high BMI significantly impaired working memory and executive function, including attention control, cognitive flexibility, inhibitory control, and task monitoring.³² However, most studies examined the effects of BMI on cognitive function in healthy people, but few explored the consequences of high BMI in AD patients. In this study, compared to AD-nBMI group, AD-hBMI group exhibited cognitive decline, with higher BMI associated with lower cognitive level in AD patients, even after adjusting for education level, smoking, drinking, constipation, water intake, history of hypertension, diabetes, and coronary heart disease (Figure 1A, B). These findings suggested that high BMI exacerbated cognitive impairment in AD patients.

Neuroinflammation was intensified and correlated with aggravated AD pathology in AD-hBMI group

To explore the specific mechanisms by which high BMI leads to cognitive decline in AD patients, an increasing number of studies have focused on neuroinflammation in brain. High-fat feeding in mice promoted neuroinflammation indicated by the over-activations of astrocytes and microglia in brain.³³ Transcriptomic studies of human cortical cells identified that neuroinflammation might be a key mechanism underlying obesity-related AD pathology.³⁴ However, no studies have yet investigated the impact of high BMI on neuroinflammation in the brain of AD patients.YKL-40 and sTREM2 levels specifically reflect the activation states of astrocytes and microglia, respectively. YKL-40, a biomarker of astrocyte activation, is encoded by the CHI3L1 gene. Elevated YKL-40 level in CSF indicated neuroinflammation featured by astroglial activation in AD patients.³⁵ TREM2 receptor is an innate immune receptor and type I transmembrane protein that promotes the transformation of normal microglia into disease-associated microglia through lipid-related pathways.³⁶ In this study, AD-hBMI group exhibited significantly reduced CSF sTREM2 level compared toAD-nBMI group (Figure 1E). However, the interpretation of CSF sTREM2 level in AD is complex, as sTREM2 is influenced by multiple factors, including age, sex, ethnicity, and TREM2 receptor genetic variants.³⁷ For instance, heterozygous carriers of the TREM2 risk variant TREM2-R47H showed significantly higher CSF sTREM2 levels than non-carriers,^38–39 and elevated CSF sTREM2 was associated with ameliorated cognitive decline in AD.⁴⁰ Therefore, the observed reduction in CSF sTREM2 level in AD-hBMI group should be interpreted cautiously, and the mechanisms by which obesity modulates TREM2-mediated microglial responses in AD warrant further investigation. We found that CSF YKL-40 level was positively correlated with P-tau181 level in AD patients, suggesting a possible link between astroglial activation and AD pathology in AD-hBMI group. In addition, CSF NO and ·OH levels in the CSF from AD-hBMI group were significantly elevated compared to that from AD-nBMI group, and CSF NO level was associated with high BMI in AD patients (Figure 1F-G). Furthermore, H₂O₂ level was positively associated with accumulated Aβ42 deposition and P-tau231 and T-tau levels in the brains of AD patients (Figure 2). This study, for the first time, investigated and demonstrated through the detection of neuroinflammatory factors in CSF of AD patients that high BMI promoted neuroinflammation and exacerbated AD pathology. The specific mechanism may be related to obesity resulting in increased lipid-laden microglia,⁴¹ which released high levels of reactive oxygen species and inflammatory factors, impaired phagocytic function, and exacerbated the progression of AD pathology.⁴²

BBB disruption was a key factor in exacerbating neuroinflammation in AD-hBMI group

To investigate the underlying mechanism of high BMI exacerbating neuroinflammation in AD patients, a growing number of studies have focused on the impact of high BMI on the integrity of BBB. BBB integrity is primarily maintained by the proteins in endothelial tight junction, including ZO-1, OCLN, and CLDN5, which play critical roles in regulating paracellular permeability. Under pathological conditions related to BBB disruption, these proteins are degraded, leading to significantly elevated levels of ZO-1, OCLN, and CLDN5 in CSF. RAGE and LRP1 are both expressed in endothelial cell membranes and play opposing roles in Aβ trafficking across BBB. RAGE mediates the influx of Aβ from blood into brain, whereas LRP1 facilitates the efflux of Aβ from brain into blood. Impaired BBB disrupts the equilibrium between RAGE and LRP1, resulting in upregulated RAGE and downregulated LRP1 in CSF. This imbalance promotes the accumulation of cerebral Aβ, contributing to the development and progression of AD. BBB disruption precedes neuroinflammation and AD pathology in brain, serving as a critical link at the early stages of AD.^43–45 Animal studies found that obesity increased BBB disruption in the hippocampus of aged mice, leading to the extravasation of plasma IgG and other components, thereby promoting neuroinflammation characterized by the intensified microglial activation and elevated levels of cytokines.¹⁷ However, the impact of high BMI on BBB disruption has not yet to be demonstrated in AD patients.

In this study, AD-hBMI group exhibited conspicuously increased ZO-1 and OCLN in CSF, among which, a greater increase in OCLN was associated with high BMI in AD patients (Figures 1 and 2). Furthermore, the levels of ZO-1, OCLN, and RAGE in CSF were positively linked to the levels of NO, H₂O₂, and IL-1β in the CSF from AD patients. In summary, BBB disruption indicated by the noteworthy loss of ZO-1 and OCLN in tight junctions might be the potential mechanism underlying the neuroinflammation observed in the brains of AD-hBMI group (Figure 2).

Gut dysbiosis was associated with BBB disruption in AD-hBMI group

In this study, the roles of gut microbiota and their metabolites were further investigated with aim of figuring out the mechanisms responsible for the disrupted BBB and subsequently exacerbated neuroinflammation in the brains of patients in AD-hBMI group.

High BMI altered gut microbial composition but not overall diversity in AD patients

The diversity of gut microbiota is crucial for the onset and progression of AD. A previous study found that gut microbiota diversity was reduced in AD patients compared to cognitively normal individuals.⁴⁶ High-fat diet-induced obesity in mouse model exhibited reduced gut microbiota richness by Chao1 index and diversity by Shannon index.²³ However, there were no reports on the effects of high BMI on microbiota diversity in AD patients. In this study, AD patients with normal and high BMI showed no significant differences in gut microbiota diversity (Figure 3). The findings suggest that high BMI may not significantly affect gut microbiota diversity in AD patients, which might be due to confounding factors, such as the pathological state of AD itself, dietary patterns, or pharmacological interventions.

Then, the composition of gut microbiota at family level was analyzed. It was found that AD-hBMI group had decreased Firmicutes and increased Bacteroidetes (Figure 4A-B). Firmicutes produce short-chain fatty acids, which have anti-inflammatory properties and maintain the integrity of tight junctions in intestinal epithelial cells and thus is pivotal for preserving normal gut barrier function. Bacteroidetes produce lipopolysaccharide (LPS), which stimulates immune system and exacerbates inflammation.^47–48 Therefore, a decreased abundance of Firmicutes and an increased abundance of Bacteroidetes might trigger or exacerbate peripheral inflammatory responses,⁴⁹ which might be one of the mechanisms through which gut microbiota cause damage to BBB.

Further, the composition of gut microbiota at genus level was examined. It was revealed that AD-hBMI group displayed a significantly increased abundance of g_Acidaminococcus. Although a previous study reported that the abundance of g_Acidaminococcus was increased and associated with peripheral inflammation in high BMI populations in Italy.⁵⁰, the current study is the first one to discover an increase in abundance of g_Acidaminococcus in the patients in AD-hBMI group. Additionally, AD-hBMI group exhibited a reduced abundance of g_Faecalibacillus, g_Corynebacterium, g_Flintibacter, and g_Mediterraneibacter (Figure 4C), among which, g_Flintibacter was found to be beneficial for cognitive improvement in AD mouse models,⁵¹ and the impact of remaining microbiota on AD-hBMI group has been rarely reported.

In this study, the network structure among gut microbiota was further explored. The gut microbiota network in AD-hBMI group exhibited a state of hyperconnectivity and dysregulation, with an enrichment of conditional pathogens, such as g_Escherichia and g_Klebsiella (Figure 5B). g_Escherichia and g_Klebsiella in gut are the significant sources of LPS, which have been confirmed as major players in the neuroinflammation related to AD.⁵² Their increased negative correlations within AD-hBMI group suggested a disruption of the ecological balance among microbial communities, with intensified competitive interactions. These findings suggest that high BMI may exacerbate neuroinflammation through gut microbiota dysbiosis in AD-hBMI group.

In this study, the functional prediction of differential gut microbiota was conducted. KEGG-enriched pathways in AD-hBMI group included significantly increased alkaloid biosynthesis (such as tropane, piperidine, pyridine, and isoquinoline), RNA transport, arginine and proline metabolism, and LPS biosynthesis (Figure 5D). MetaCyc pathways analyses in AD-hBMI group indicated that nucleotide, amino acid, and lipid metabolism were significantly enhanced, while energy and carbohydrate metabolism were significantly reduced (Figure 5C). Tropane has been confirmed to induce AD symptoms in animal models.⁵³ LPS was related to neuroinflammation in AD mice.⁵⁴ Enhanced nucleotide, amino acid, and lipid metabolism, and reduced energy and carbohydrate metabolism were involved metabolism and bioenergetics, which were within the Common Alzheimer's Disease Research Ontology (CADRO) categories.⁵⁵

The correlations of gut metabolites with BBB disruption and neuroinflammation in AD-hBMI group

In this study, the impact of high BMI on gut metabolites in AD patients was further explored. It was observed that the increased metabolites in AD-hBMI group included hormones, reductants, amino acids and lipid metabolism, and neuro-regulators. The reduced metabolites in AD-hBMI group included bile acid metabolism, reductants, and estrogen (Figure 6B), which might exacerbate AD pathology by influencing neurological regulators, oxidative stress, and metabolism according to CADRO categories.⁵⁵

Furthermore, differences in metabolites associated with BBB disruption and neuroinflammation between AD-hBMI and AD-nBMI groups were identified. Firstly, we found that the following altered gut metabolites were significantly correlated with BBB disruption indicaated by the elevated ZO-1 and OCLN levels in the CSF from AD-hBMI group (Figure 7B). In AD-hBMI group, elevated thioridazine, 4-hydroxybenzaldehyde, and N-acetylleucine were positively correlated with ZO-1 level in CSF, reduced acesulfame and undecanoic acid were negatively correlated with OCLN level in CSF, and elevated (S)-S-methylcysteine sulfoxide was positiveiy correlated with both ZO-1 and OCLN levels in CSF. These findings suggested that the levels of the aforementioned metabolites were closely associated with the integrity of tight junctions, thus exacerbating AD progression by increasing BBB permeability in AD-hBMI group. Secondly, we observed that the following altered differential gut metabolites were significantly associated with neuroinflammation featured by the increased NO and ·OH levels in the CSF from AD-hBMI group. The increased levels of gut thioridazine and medroxyprogesterone were correlated with NO level in the CSF from AD-hBMI group, which might elicit neuroinflammation and neuronal damage in AD-hBMI patients as it did in stroke-induced rats.⁵⁶ Meanwhile, the decreased butyric acid level was correlated with the elevated ·OH level in the CSF from AD-hBMI group, implying that butyric acid might reduce oxidative stress through anti-inflammatory and neuroprotective effects in AD-hBMI patients as it did in AD mice.⁵⁷ These findings suggested that BBB disruption and neuroinflammation might impact brain through gut-brain axis mechanisms in AD-hBMI patients.

In summary, AD-hBMI patients exhibit impaired overall cognition, disrupted BBB (elevated ZO-1 and OCLN levels in CSF), and intensified neuroinflammation (elevated NO and ·OH levels in CSF) compared with AD-nBMI patients. BMI is positively correlated with cognitive impairment, OCLN and NO levels in CSF, while ZO-1 is positively correlated with NO levels in the CSF from AD patients. AD-hBMI patients display a unique gut dysbiosis pattern characterized by the alterations in specific metabolite levels. In AD-hBMI group, elevated thioridazine, 4-hydroxybenzaldehyde, and N-acetylleucine were positively correlated with ZO-1 level in CSF, reduced acesulfame and undecanoic acid were negatively correlated with OCLN level in CSF, and elevated (S)-S-methylcysteine sulfoxide was positiveiy correlated with both ZO-1 and OCLN levels in CSF. Elevated levels of thioridazine and medroxyprogesterone are correlated with neuroinflammationindicated by an increased of NO level in CSF. The decreased butyric acid was associated with neuroinflammation reflected by the elevated ·OH level in CSF. These findings suggest that gut dysbiosis may exacerbate BBB disruption and neuroinflammation via gut-brain axis in AD-hBMI patients.

This study has several limitations. First, the relatively small sample size limits the generalizability of these findings. Future multicenter studies with larger cohorts are needed to validate the robustness and applicability of these results. Second, the cross-sectional design of this study precludes establishing causal relationships among gut microbiota dysbiosis, BBB disruption, neuroinflammation, and clinical symptoms in AD-hBMI patients. Longitudinal studies with larger sample sizes are necessary to elucidate these causal links and reveal whether the rate of cognitive decline of AD-hBMI patients is faster than that of AD-nBMI patients. Third, potential confounding factors, such as diet and medication, were not investigated, which may influence the observed associations. Comprehensive analyses in the future studies are required to confirm and extend these findings.

Conclusion

High BMI may accelerate gut dysbiosis, disrupt BBB, intensify neuroinflammation, and exacerbate cognitive decline in AD patients. Findings from this investigation may provide perspectives on the novel mechanisms and interventional strategies for AD.

Footnotes

Acknowledgements

We thank all participants involved in the present study.

ORCID iD

Wei Zhang

Ethical considerations

This investigation received ethical clearance from the Institutional Review Board of Beijing Tiantan Hospital affiliated to Capital Medical University (KY2024-260-03).

Consent to participate

In line with the Declaration of Helsinki, written informed consent forms were duly signed by all the participants and their family members enrolled in this study.

Consent for publication

Not applicable

Author contribution(s)

Jing Li: Investigation; Writing – original draft; Writing – review & editing.

Tenghong Lian: Data curation; Writing – original draft.

Peng Guo: Formal analysis; Writing – original draft.

Jinghui Li: Writing – review & editing.

Jing Qi: Methodology; Writing – original draft.

Dongmei Luo: Resources; Writing – original draft.

Yao Meng: Software; Writing – original draft.

Mingyue He: Software; Writing – original draft.

Zijing Zheng: Supervision; Writing – original draft.

Hao Yue: Supervision; Writing – original draft.

Zhan Liu: Validation; Writing – original draft.

Fan Zhang: Data curation.

Ruidan Wang: Data curation.

Weijia Zhang: Data curation.

Wei Zhang: Funding acquisition; Project administration; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Study on the Special Project on New Technology-Medical Integration Innovation under the China University Industry-Academia-Research Innovation Fund (grant number 2025XJS014), Capital's Funds for Health Improvement and Research (CFH) (grant number 2022-2-2048), Collaborative Research Project of Traditional Chinese and Western Medicine of the Major Difficult Disease-Alzheimer's Disease of Beijing (grant number 2023BJSZDYNJBXTGG-018), Clinical Collaboration Project on Major and Difficult Diseases with Integrated Traditional Chinese and Western Medicine: Vascular Dementia (grant number No. ZDYN-2024-A-008), the National Natural Science Foundation of China (grant number 82501824, 81970992), the National Key Research and Development Program of China (grant number 2016YFC1306300), STI2030-Major Projects Youth Scientist Program (grant number 2022ZD0213600), Science and Technology Innovation 2030 Major Projects (grant number 2022ZD0211600), the Research on Mechano-Biomaterial Sciences in Brain Diseases and Neuromodulation (grant number T2488101), the Project of Scientific and Technological Development of Traditional Chinese Medicine in Beijing (grant number JJ2018-48), the Natural Science Foundation of Hebei, China (grant number H2021206416), Medical Science Research Project of Health Commission of Hebei, China (grant number 20221377).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets presented in this study can be found in online repositories. Accession number(s): NCBI, PRJNA1196923.

Supplemental material

Supplemental material for this article is available online.

References

Alzheimer's Association. 2021 Alzheimer's disease facts and figures. Alzheimer's Dement 2021; 17: 327–406.

Long

Holtzman

. Alzheimer disease: an update on pathobiology and treatment strategies. Cell 2019; 179: 312–339.

Bellenguez

Küçükali

Jansen

, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet 2022; 54: 412–436.

Barnes

Yaffe

. The projected effect of risk factor reduction on Alzheimer's disease prevalence. Lancet Neurol 2011; 10: 819–828.

Bischof

Park

. Obesity and aging: consequences for cognition, brain structure, and brain function. Psychosom Med 2015; 77: 697–709.

Dye

Boyle

Champ

, et al. The relationship between obesity and cognitive health and decline. Proc Nutr Soc 2017; 76: 443–454.

Woodie

Blythe

. The differential effects of high-fat and high-fructose diets on physiology and behavior in male rats. Nutr Neurosci 2018; 21: 328–336.

Mengr

Strnadová

Strnad

, et al. Feeding high-fat diet accelerates development of peripheral and central insulin resistance and inflammation and worsens AD-like pathology in APP/PS1 mice. Nutrients 2023; 15: 3690.

Whitmer

Gustafson

Barrett-Connor

, et al. Central obesity and increased risk of dementia more than three decades later. Neurology 2008; 71: 1057–1064.

10.

Jagust

Harvey

Mungas

, et al. Central obesity and the aging brain. Arch Neurol 2005; 62: 1545–1548.

11.

Balasubramanian

Kiss

Tarantini

, et al. Obesity-induced cognitive impairment in older adults: a microvascular perspective. Am J Physiol Heart Circ Physiol 2021; 320: H740–H761.

12.

Buie

Watson

Smith

, et al. Obesity-related cognitive impairment: the role of endothelial dysfunction. Neurobiol Dis 2019; 132: 104580.

13.

de Paula

Brunetta

Engel

, et al. Hippocampal function is impaired by a short-term high-fat diet in mice: increased blood-brain barrier permeability and neuroinflammation as triggering events. Front Neurosci 2021; 15: 734158.

14.

Haseloff

Dithmer

Winkler

, et al. Transmembrane proteins of the tight junctions at the blood-brain barrier: structural and functional aspects. Semin Cell Dev Biol 2015; 38: 16–25.

15.

Jiao

, et al. The role of circulating tight junction proteins in evaluating blood brain barrier disruption following intracranial hemorrhage. Dis Markers 2015; 2015: 860120.

16.

Zhu

, et al. Circulating tight junction proteins mirror blood-brain barrier integrity in leukaemia central nervous system metastasis. Hematol Oncol 2017; 35: 365–373.

17.

Tucsek

Toth

Sosnowska

, et al. Obesity in aging exacerbates blood-brain barrier disruption, neuroinflammation, and oxidative stress in the mouse hippocampus: effects on expression of genes involved in beta-amyloid generation and Alzheimer's disease. J Gerontol A Biol Sci Med Sci 2014; 69: 1212–1226.

18.

Pallebage-Gamarallage

Lam

Takechi

, et al. Restoration of dietary-fat induced blood-brain barrier dysfunction by anti-inflammatory lipid-modulating agents. Lipids Health Dis 2012; 11: 117.

19.

Neyra Chauca

Robles Martinez

. Histological and functional breakdown of the blood-brain barrier in Alzheimer's disease: a multifactorial intersection. Neurol Int 2025; 17: 166.

20.

Cryan

O'Riordan

Cowan

CSM

, et al. The microbiota-gut-brain axis. Physiol Rev 2019; 99: 1877–2013.

21.

Donoso

Cryan

Olavarría-Ramírez

, et al. Inflammation, lifestyle factors, and the microbiome-gut-brain axis: relevance to depression and antidepressant action. Clin Pharmacol Ther 2023; 113: 246–259.

22.

Shi

Lin

, et al. β-glucan attenuates cognitive impairment via the gut-brain axis in diet-induced obese mice. Microbiome 2020; 8: 143.

23.

Shi

Wang

Zheng

, et al. Supplement of microbiota-accessible carbohydrates prevents neuroinflammation and cognitive decline by improving the gut microbiota-brain axis in diet-induced obese mice. J Neuroinflammation 2020; 17: 77.

24.

Sherwin

Bordenstein

Quinn

, et al. Microbiota and the social brain. Science 2019; 366: eaar2016.

25.

Chen

Liao

Xia

, et al. Gut microbiota regulate Alzheimer's disease pathologies and cognitive disorders via PUFA-associated neuroinflammation. Gut 2022; 71: 2233–2252.

26.

Jack

Jr Bennett

Blennow

, et al. NIA-AA research framework: toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018; 14: 535–562.

27.

Zhou

. Predictive values of body mass index and waist circumference for risk factors of certain related diseases in Chinese adults–study on optimal cut-off points of body mass index and waist circumference in Chinese adults. Biomed Environ Sci 2002; 15: 83–96.

28.

Cockrell

Folstein

. Mini-mental state examination (MMSE). Psychopharmacol Bull 1988; 24: 689–692.

29.

Huang

. The Beijing version of the Montreal Cognitive Assessment as a brief screening tool for mild cognitive impairment: a community-based study. BMC Psychiatry 2012; 12: 156.

30.

Luchsinger

Gustafson

. Adiposity and Alzheimer's disease. Curr Opin Clin Nutr Metab Care 2009; 12: 15–21.

31.

Zhang

Chen

. Body mass index and trajectories of the cognition among Chinese middle and old-aged adults. BMC Geriatr 2022; 22: 613.

32.

Yang

Shields

Guo

, et al. Executive function performance in obesity and overweight individuals: a meta-analysis and review. Neurosci Biobehav Rev 2018; 84: 225–244.

33.

Lin

Basu

Chatterjee

, et al. Disease-associated astrocytes and microglia markers are upregulated in mice fed high fat diet. Sci Rep 2023; 13: 12919.

34.

Patel

Woo

Shi

, et al. Obesity and the cerebral cortex: underlying neurobiology in mice and humans. Brain Behav Immun 2024; 119: 637–647.

35.

Lananna

McKee

King

, et al. Chi3l1/YKL-40 is controlled by the astrocyte circadian clock and regulates neuroinflammation and Alzheimer's disease pathogenesis. Sci Transl Med 2020; 12: eaax3519.

36.

Keren-Shaul

Spinrad

Weiner

, et al. A unique microglia type associated with restricting development of Alzheimer's disease. Cell 2017; 169: 1276–1290.e17.

37.

Filipello

You

Mirfakhar

, et al. Defects in lysosomal function and lipid metabolism in human microglia harboring a TREM2 loss of function mutation. Acta Neuropathol 2023; 145: 749–772.

38.

Piccio

Deming

Del-Aguila

, et al. Cerebrospinal fluid soluble TREM2 is higher in Alzheimer disease and associated with mutation status. Acta Neuropathol 2016; 131: 925–933.

39.

Suárez-Calvet

Morenas-Rodríguez

Kleinberger

, et al. Early increase of CSF sTREM2 in Alzheimer's disease is associated with tau related-neurodegeneration but not with amyloid- β pathology. Mol Neurodegener 2019; 14: 1.

40.

Ewers

Franzmeier

Suárez-Calvet

, et al. Increased CSF soluble TREM2 is associated with ameliorated longitudinal cognitive and clinical decline in Alzheimer's disease. Alzheimers Dement 2019; 15: P917.

41.

Liang

Gong

Zhao

, et al. Long-term high-fat diet consumption aggravates β-amyloid deposition and tau pathology accompanied by microglial activation in an Alzheimer's disease model. Mol Nutr Food Res 2024; 68: e2300669.

42.

Marschallinger

Iram

Zardeneta

, et al. Lipid-droplet-accumulating microglia represent a dysfunctional and proinflammatory state in the aging brain. Nat Neurosci 2020; 23: 194–208.

43.

Alkhalifa

Al-Ghraiybah

Odum

, et al. Blood-brain barrier breakdown in Alzheimer's disease: mechanisms and targeted strategies. Int J Mol Sci 2023; 24: 16288.

44.

Kalaria

. Vascular basis for brain degeneration: faltering controls and risk factors for dementia. Nutr Rev 2010; 68: S74–S87.

45.

Karczewski

Zielińska

Staszewski

, et al. Obesity and the brain. Int J Mol Sci 2022; 23: 6145.

46.

Liu

Peng

, et al. Altered microbiomes distinguish Alzheimer's disease from amnestic mild cognitive impairment and health in a Chinese cohort. Brain Behav Immun 2019; 80: 633–643.

47.

Lassenius

Pietiläinen

Kaartinen

, et al. Bacterial endotoxin activity in human serum is associated with dyslipidemia, insulin resistance, obesity, and chronic inflammation. Diabetes Care 2011; 34: 1809–1815.

48.

Wang

. The relationship between gut microbiota and inflammatory diseases: the role of macrophages. Front Microbiol 2020; 11: 1065.

49.

Chandra

Sisodia

Vassar

. The gut microbiome in Alzheimer's disease: what we know and what remains to be explored. Mol Neurodegener 2023; 18: 9.

50.

Palmas

Pisanu

Madau

, et al. Gut microbiota markers associated with obesity and overweight in Italian adults. Sci Rep 2021; 11: 5532.

51.

Lamichhane

Liu

Lee

, et al. Curcumin mitigates the high-fat high-sugar diet-induced impairment of spatial memory, hepatic metabolism, and the alteration of the gut microbiome in Alzheimer's disease-induced (3xTg-AD) mice. Nutrients 2024; 16: 261.

52.

Zhao

Jaber

Pogue

, et al. Lipopolysaccharides (LPSs) as potent neurotoxic glycolipids in Alzheimer's disease (AD). Int J Mol Sci 2022; 23: 12436.

53.

Zhao

, et al. Anti-oxidative stress and cognitive improvement of a semi-synthetic isoorientin-based GSK-3β inhibitor in rat pheochromocytoma cell PC12 and scopolamine-induced AD model mice via AKT/GSK-3β/Nrf2 pathway. Exp Neurol 2024; 380: 114881.

54.

Yang

García-Cruzado

Zeng

, et al. LPS Priming before plaque deposition impedes microglial activation and restrains Aβ pathology in the 5xFAD mouse model of Alzheimer's disease. Brain Behav Immun 2023; 113: 228–247.

55.

Refolo

Snyder

Liggins

, et al. Common Alzheimer's disease research ontology: national institute on aging and Alzheimer's association collaborative project. Alzheimers Dement 2012; 8: 372–375.

56.

Tang

Chen

, et al. Paeoniflorin improves functional recovery through repressing neuroinflammation and facilitating neurogenesis in rat stroke model. PeerJ 2021; 9: e10921.

57.

Bai

Jiang

, et al. Butyrate alleviates PTZ-induced mitochondrial dysfunction, oxidative stress and neuron apoptosis in mice via Keap1/Nrf2/HO-1 pathway. Brain Res Bull 2021; 168: 25–35.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.36 MB

0.00 MB