Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Abstract

Background

Neuroinflammation represents a central pathological mechanism in Alzheimer's disease (AD). Lipopolysaccharide (LPS) is a potent inducer of neuroinflammation and demonstrates elevated circulating levels in AD patients.

Objective

This study aims to investigate the genetic association between serum LPS activity level, inflammatory proteins and AD.

Methods

A two-sample mendelian randomization (MR) analysis was performed to explore the causal effect of serum LPS activity level and 91 inflammatory proteins on AD, including 1, 260, 136 sporadic AD and 2, 838, 825 familial AD patients, respectively. Meta-analysis was conducted on multiple datasets to determine statistically significant results that was initially observed in one dataset.

Results

Serum LPS activity level is a risk factor for early onset sporadic AD with OR = 1.392, 95% CI: 1.038–1.869. In most other sporadic AD datasets, LPS shows a trend of increasing the risk of AD onset. After meta-analysis in 10 independent datasets, no association between LPS and sporadic AD was observed. In most familial AD datasets, LPS level demonstrated a trend of decreasing AD risk in MR analysis, however, meta-analysis of the combined 8 datasets showed no statistically significant difference. Two inflammatory proteins, AXIN1 and IL-1 alpha, were identified as significant risk factors for sporadic AD.

Conclusions

This study suggested that serum LPS activity level may present a risk effect in early onset sporadic AD. Two inflammatory proteins AXIN1 and IL-1 alpha were associated with the risk of sporadic AD. These findings provide a new perspective for the early diagnosis and treatment of sporadic and familial AD.

Keywords

Alzheimer's disease inflammatory proteins lipopolysaccharide Mendelian randomization meta-analysis

Introduction

Alzheimer's disease (AD), recognized as the leading cause of neurodegenerative disease, is marked by increase of hyperphosphorylated tau (pTau) and aggregation of amyloid-β (Aβ).¹ AD spans a continuum from cognitively normal to mild cognitive impairment and dementia, emphasizing AD progression over several years.² Statistics reveal that approximately one in nine individuals aged 65 and older, or 10.8%, suffer from AD in Americans.³ In 2019, AD was recognized as the sixth leading cause of death in the United States, with 121,499 fatalities attributed to the condition.⁴ It was estimated, the prevalence of AD will reach three times higher worldwide by 2050 when diagnosing based on biological indicators.² These data underscore the profound impact of this neurodegenerative disorder on the aging demographic. However, the pathogenesis underlying AD are not fully understood.

Risk factors of AD encompass age, genetic mutations or variants, traumatic brain injury, diabetes and infections.¹ AD can be divided into sporadic and familial types, accounting for 95% and 5% respectively. Sporadic AD with an age of onset at 60–65 years account 90% AD are late-onset AD (LOAD).⁵ APOE4 allele is a strongest genetic risk factor of LOAD, which can promote Aβ aggregation and affect tau pathogenesis.^6,7 Familial AD is primarily caused by mutations of amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2), and mutation carriers exhibited higher levels of Aβ deposition, decreased metabolism, and cortical thinning compared to non-carriers.⁸ In addition, genome-wide association studies (GWAS) have identified plenty of AD associated genes, which can increase the AD susceptibility, indicating the critical role of genetic factors in the onset and progression of AD.^9–12

Brain infection is another major risk factor for AD.¹ Persistent neuroinflammation triggered by various infectious agents can facilitate the initiation and advancement of AD.¹³ Bacterial such as porphyromonas gingivalis, can harm cells utilizing their lipopolysaccharides (LPS) by combining with Toll-like receptors (TLRs), thereby inducing the secretion of pro-inflammatory cytokines to damage the neurons.¹⁴ Activation of microglia and astrocytes are involved in both neuroprotective (M2-phenotype microglia and A2-phenotype astrocytes) and neurotoxic (M1-phenotype microglia and A1-phenotype astrocytes) functions.^1,15 During the development of AD, the continuous chronic activation of microglia by Aβ will lead to a long-lived inflammation through producing extensive cytokines, thereby promoting the progress of AD.¹⁶ A1-type astrocytes activated by LPS and IFN-γ show a decrease in phagocytic ability and dysregulation of synaptic function.¹⁵ Additionally, pro-inflammatory microglia can activate pro-inflammatory astrocytes by releasing inflammatory factors such as IL-1α, IL-1β, TNF-α, result in a secondary inflammatory response.¹⁵ Therefore, targeting the pathways driven by microglia and astrocytes could inhibit neuroinflammation to intervene the progression of AD.¹⁷

Neuroinflammation is one of the characteristics of brain tissue in AD.² LPS, commonly referred to as endotoxin, is a significant virulence factor of a cell-envelope component from gram-negative bacteria. Studies reported that LPS endotoxin levels in blood are higher in AD patients than in normal individuals.^18,19 LPS can induce neuroinflammation in the brain, and the immunomodulatory itaconate can reduce the inflammation by downregulating levels of inflammatory mediators.²⁰ Neuroinflammatory biomarkers have the potential to achieve personalized treatment for AD by anti-neuroinflammatory therapy.²¹ However, detecting early-stage AD in clinical settings is difficult, as symptoms are frequently overlooked as routine aspects of aging by both patients and healthcare providers. Therefore, it is necessity to screen and identify potential novel biomarkers in AD's earliest stages in order to promote disease improvement and the development of secondary prevention therapies before symptoms appear.²² It is well established that LPS can induce neuroinflammation; however, there is a lack of research on whether populations susceptible to blood LPS level are at an increased risk for neuroinflammation-related diseases such as AD. Furthermore, the causal relationship between the multitude of inflammatory factors implicated in the pathogenesis of AD and the disease itself remains poorly understood.

Mendelian randomization (MR) studies can estimate the unconfounded causal effect.²³ In this study, a two-sample reverse MR analysis was employed to elucidate the causal relationship between LPS/inflammatory factors and AD. Our research is poised to facilitate advancements in early detection, accelerate drug discovery, and refine personalized therapeutic strategies for both sporadic and familial AD.

Methods

Data availability and sources

All data used in this study are publicly available from GWAS databases. LPS related dataset (ebi-a-GCST90032674: the quantification of LPS activity in serum) was obtained from the Integrative Epidemiology Unit (IEU) OpenGWAS database (https://gwas.mrcieu.ac.uk/datasets/).

The sporadic AD related 10 datasets in this study were from two databases: the GWAS catalog database (https://www.ebi.ac.uk/gwas/) including five datasets: ebi-a-GCST002245 (late onset), ebi-a-GCST90027158, ieu-a-824, ieu-b-2. ieu-b-5067 and the Finnish database (https://www.finngen.fi/en/access_results) including five datasets: finngen_R10_AD_AM_EXMORE (atypical or mixed), finngen_R10_AD_EO_EXMORE (early onset), finngen_R10_AD_LO_EXMORE (late onset), finngen_R10_AD_U_EXMORE, and finngen_R10_G6_ALZHEIMER.

The datasets related to AD with a family history was drawn from the IEU OpenGWAS database including eight datasets: ebi-a-GCST005920, ebi-a-GCST005921, ebi-a-GCST005923, ebi-a-GCST90012877, ebi-a-GCST90012878, ukb-a-210, ukb-b-14043, ukb-b-14699.

Inflammatory factor-related datasets were obtained from the GWAS catalog database including dataset from GCST90274758 to GCST90274848 [PMID:37563310]. Plasma levels of 91 inflammatory proteins in European populations (Sweden, Denmark, United Kingdom, Germany, Estonia and Croatia) were tested using genome-wide genotyping array technology.

Ethical approval was not required for this study, as we used the results of publicly available GWAS analyses. The information of these datasets was list in Table 1.

Table 1.

Information of the exposure and outcome datasets in this study.

Variable	Source	Ethnicity	Sample size
Exposure
Serum lipopolysaccharide activity	ebi-a-GCST90032674	European	11,296
Inflammatory factor	GCST90274758–GCST90274848(91)	European	1585
Outcome
Alzheimer's disease (late onset)	ebi-a-GCST002245	European	55,134
Alzheimer's disease	ebi-a-GCST90027158	European	487,511
Alzheimer's disease	ieu-a-824	European	1489
Alzheimer's disease	ieu-b-2	European	63,926
Alzheimer's disease	ieu-b-5067	European	488,285
Alzheimer disease (Atypical or mixed)	finngen_R10_AD_AM_EXMORE	European	25,390
Alzheimer disease (Early onset)	finngen_R10_AD_EO_EXMORE	European	16,730
Alzheimer disease (Late onset)	finngen_R10_AD_LO_EXMORE	European	26,920
Alzheimer disease (undefined)	finngen_R10_AD_U_EXMORE	European	5062
Alzheimer disease	finngen_R10_G6_ALZHEIMER	European	89,689
Paternal history of Alzheimer's disease	ebi-a-GCST005920	European	260,279
Family history of Alzheimer's disease	ebi-a-GCST005921	European	314,278
Maternal history of Alzheimer's disease	ebi-a-GCST005923	European	288,676
Alzheimer's disease or family history of Alzheimer's disease	ebi-a-GCST90012877	European	472,868
Family history of Alzheimer's disease	ebi-a-GCST90012878	European	408,942
Illnesses of mother: Alzheimer's disease/dementia	ukb-a-210	European	308,780
Illnesses of siblings: Alzheimer's disease/dementia	ukb-b-14043	European	361,264
Illnesses of mother: Alzheimer's disease/dementia	ukb-b-14699	European	423,738

Selection of genetic instrumental variables

Single nucleotide polymorphisms (SNPs) are used as instrumental variables (IVs) for exposure and outcome. A relatively rigorous screening criteria was used to identify eligible SNPs for subsequent analyses. First, SNPs should be strongly associated with exposures with genome-wide significance levels of p < 1 × 10⁻⁵. Second, to avoid biased results due to linkage disequilibrium (LD), the minimal linkage disequilibrium (r² < 0.001) were identified within the 10,000 kb range from GWAS, SNPs in the range of 10,000 kb with r²> 0.001 were filtered out. Third, IVs with F value < 10 were defined as weak IVs and were removed for subsequent analysis. For IVs of serum LPS levels selection, when clump_kb = 5000, clump_r2 = 0.01 and p < 1 × 10⁻⁵ runs, only one SNP reach the specification. To increase the number of SNPs, IVs of serum LPS levels were selected by using p < 1 × 10⁻⁵ and F value > 10 criterion.

Statistical analysis

Mendelian randomization analysis

In this study, we used five MR methods to estimate the causal effect of exposure on outcome, namely MR Egger regression, Weighted median, Inverse-Variance Weighted MR, Simple mode and Weighted mode. “TwoSampleMR” package, “VariantAnnotation” package and “gwasglue” package of R (version 4.3.2) were performed for two-sample MR analysis. Among them, IVW is the most widely used method in MR analysis, which can provide a more accurate estimation of causal effects through inverse variance weighting. In our study, a p-value < 0.05 of IVW method was considered have statistical differences. We used a leave-one-out sensitivity, MR-Egger and Cochran's Q test analysis to assess whether the results were potentially biased by genetic pleiotropy and data heterogeneity. A p-value >0.05 was considered no genetic pleiotropy and data heterogeneity. When there is a contradiction between the result of MR-Egger and IVW methods, MR-Egger and Cochran's Q test were used to determine which result was adopted. If the intercept term of MR Egger intercept test is significant (p-value < 0.05), indicating the existence of directional pleiotropy, IVW results may be biased, and we prioritized the MR Egger result as the primary finding. Conversely, both Cochran's Q test and MR Egger intercept test are not significant (p-value > 0.05), IVW results were used.

FDR correction

The False Discovery Rate (FDR) is the proportion of multiple hypothesis tests that are misreported as true associations. “pacman” package was used for FDR adjustment in MR analysis between inflammatory proteins and AD. All statistically significant associations were FDR corrected with adjusted p-values < 0.05.

Meta-analysis

Since the outcome data in this study was from different datasets, in order to estimate causality more accurately and improve the statistical power, we integrate the results by performing a meta-analysis using “meta” package. By combining meta-analysis and MR methods, the associations between genotypes and phenotypes were explored more comprehensively to improve the credibility and accuracy of this paper. Furthermore, we performed funnel plot analysis to evaluate potential bias in the MR findings obtained from the different AD datasets.

Results

Overview of this study

Figure 1 summarized the simple study flow and results. In this study, the serum LPS activity level and 91 inflammation proteins were used as exposure factors, sporadic AD and familial AD were used as outcome to conduct a two-sample MR analysis, respectively. For the positive results in any dataset, in order to increase statistical power value, meta-analysis was performed in all sporadic AD or familial AD patients. Our results indicate that LPS levels have risk effect on early onset sporadic AD, and there was no causal effect of LPS levels on both sporadic AD and familial AD patients after meta-analysis. Two inflammatory factors AXIN1 and IL1 have casual effect on risk of sporadic AD, and no inflammatory factors have association with familial AD after meta-analysis.

Figure 1.

Flowchart of the study. The serum LPS activity level and 91 inflammation proteins were used as exposure factors, sporadic AD and familial AD were used as outcome to conduct a two-sample MR analysis, respectively. After heterogeneity and pleiotropy test, meta-analysis was performed. LPS levels have risk effect on early onset sporadic AD, and there was no causal effect of LPS levels on both sporadic AD and familial AD patients after meta-analysis. 17 inflammatory factors were identified have causal effect on sporadic AD. Two inflammatory factors AXIN1 and IL1 have casual effect on risk of sporadic AD, and no inflammatory factors have association with familial AD after meta-analysis.

Causal effect of LPS on sporadic AD

After removing weak IVs (F < 10), the final number of SNPs associated with the serum LPS activity level with p-value < 1 × 10⁻⁵ was 71 (Supplemental Table 1). The MR result showed there were statistical differences in five datasets (finngen_R10_AD_EO_EXMORE, finngen_R10_G6_AD_WIDE, finngen_R10_G6_ALZHEIMER, ebi-a-GCST90027158, and ieu-b-5067) with p-value < 0.05 using IVW method, however, no dataset with same OR value direction in all five MR methods (Figure 2). We determined whether the IVW or MR Egger results could be used for the next meta-analysis by using sensitivity analysis including heterogeneity and pleiotropy tests (Table 2). And the result of MR Egger with OR = 1.392 (95%CI: 1.038–1.869) indicated that serum LPS levels were a risk factor for early-onset AD. Supplemental Figure 1 showed the forest plot, scatter plot, funnel plot and leave-one-out analysis of LPS level IVs with causal effect on early-onset sporadic AD (finngen_R10_AD_EO_EXMORE). We then performed a meta-analysis using IVW or Egger results selected in the previous step in all outcome datasets. The meta-analysis result of heterogeneity test showed that the p value > 0.05 and I² = 18.1%, which means there is not a significant heterogeneity among these data. The result of common effects model (OR = 1.000, 95% CI: 1.000–1.001) is used. That is, LPS activity level didn’t exhibit a causal effect on sporadic AD (Figure 3). The detailed MR results of LPS activity level in sporadic AD were list in the Supplemental Table 2.

Figure 2.

The effect of the serum LPS activity levels on sporadic AD. Effect of serum LPS activity levels on sporadic AD in 10 different outcome datasets. nsnp: number of single nucleotide polymorphisms; OR: odds ratio; CI: confidence interval; FDR: false discovery rate.

Figure 3.

Meta-analysis result of MR results between serum LPS activity levels and sporadic AD. Above is a forest chart and below is a funnel chart to eliminate data bias.

Table 2.

Heterogeneity and pleiotropy analysis of MR analysis between LPS and AD, and the MR results selection.

AD datasets	Heterogenity		MR-Egger intercept			MR result
	Q	Q-pvalue	Egger_intercept	SE	p	OR(95% CI)	p	result selection
ebi-a-GCST002245	13.229	1.000	−0.004	0.005	0.390	1.004 (0.940 to 1.073)	0.896	Inverse variance weighted
ebi-a-GCST90027158	681.749	0.485	−0.006	0.002	0.012	1.012 (0.963 to 1.063)	0.641	MR Egger
ieu-a-824	0.136	0.998	0.003	1.128	0.998	1.053 (0.320 to 3.466)	0.932	Inverse variance weighted
ieu-b-2	31.850	0.965	0.004	0.004	0.382	0.984 (0.929 to 1.043)	0.984	Inverse variance weighted
ieu-b-5067	37.684	0.964	0.000	0.000	0.343	1.000 (1.000 to 1.001)	0 . 035	Inverse variance weighted
finngen_R10_AD_AM_EXMORE	42.156	0.952	−0.020	0.011	0.082	0.935 (0.820 to 1.065)	0.312	Inverse variance weighted
finngen_R10_AD_EO_EXMORE	50.446	0.749	−0.065	0.014	0.000	1.392 (1.038 to 1.869)	0 . 031	MR Egger
finngen_R10_AD_LO_EXMORE	74.026	0.090	−0.021	0.007	0.005	0.917 (0.837 to 1.004)	0.061	Inverse variance weighted
finngen_R10_AD_U_EXMORE	41.938	0. 944	−0.077	0.025	0.003	1.534 (0.895 to 2.630)	0.125	MR Egger
finngen_R10_G6_ALZHEIMER	29.193	0.999	−0.015	0.005	0.006	1.035 (0.921 to 1.162)	0.567	MR Egger
ebi-a-GCST005920	55.444	0.245	0.006	0.003	0..053	1.013 (0.986 to 1.040)	0.348	Inverse variance weighted
ebi-a-GCST005921	53.832	0.261	0.007	0.002	0.002	0.991 (0.971 to 1.011)	0.382	MR Egger
ebi-a-GCST005923	37.543	0.861	0.007	0.002	0.006	0.991 (0.967 to 1.015)	0.470	MR Egger
ebi-a-GCST90012877	38.122	0.974	0.008	0.003	0.004	0.952 (0.898 to 1.008)	0.099	MR Egger
ebi-a-GCST90012878	4.462	0.276	0.001	0.000	0.006	0.997 (0.992 to 1.001)	0.152	MR Egger
ukb-a-210	60.529	0.252	0.000	0.000	0.048	1.001 (0.996 to 1.005)	0.815	MR Egger
ukb-b-14043	9.599	0.998	0.000	0.000	0.239	0.998 (0.997 to 1.000)	0 . 011	Inverse variance weighted
ukb-b-14699	41.296	0.898	0.000	0.000	0.246	1.001 (0.999 to 1.004)	0.227	Inverse variance weighted

Causal effect of LPS on familial AD

We also performed MR analysis with LPS as the exposure factor and familial AD as the outcome. Figure 4 showed there were 2 datasets (ukb-b-14043 and ukb-a-210) with p value < 0.05 in the IVW method. Among them, there was only ukb-a-210 with the consistent direction of OR value in 5 MR methods, and the OR value was 1.004 (95% CI: 1.000–1.007, p = 0.003), however, there was pleiotropy in ukb-a-210 (p-value < 0.05 of egger intercept test). Meta-analysis was conducted after heterogeneity and pleiotropy tests (Table 2). Meta-analysis results showed that p value >0.05, I² = 38.3% (I² > 50%), suggesting no heterogeneity among these datasets. The common effect model was adopted with OR = 0.999, 95% CI: 0.998–1.000, no statistical differences were observed (Figure 5). Therefore, LPS activity level were not considered to be a risk factor for familial AD. However, LPS activity level showed the opposite effects in sporadic (potential risk effect) and familial AD (potential protective effects). The detailed MR results of LPS activity level in familial AD were list in the Supplemental Table 3.

Figure 4.

The effect of the serum LPS activity levels on familial AD. Effect of serum LPS activity levels on familial AD in 8 different outcome datasets. nsnp: number of single nucleotide polymorphisms; OR: odds ratio; CI: confidence interval; FDR: false discovery rate.

Figure 5.

Meta-analysis result of MR results between serum LPS activity levels and familial AD. Above is a forest chart and below is a funnel chart to eliminate data bias.

Causal effect of 91 inflammatory proteins on sporadic AD

Chronic inflammation is one of the risk factors for the onset of AD. Therefore, we simultaneously explored the causal relationship between 91 inflammatory proteins and AD. After removing SNPs exist LD (kb = 1000, r² = 0.001) and weak IVs (F < 10), the final number of SNPs associated with 91 inflammatory proteins with p-value <1 × 10⁻⁵ was 2726. Supplemental Table 4 showed the information of selected 2726 IVs of 91 inflammatory proteins. The results showed, after FDR correction of p-value, 10 inflammatory proteins (ARTN, AXIN1, CX3CL1, CXCL9, IFN-gamma, IL-1 alpha, IL5, SIRT2, TNFSF14 and TSLP) were associated with the risk of sporadic AD with OR value from 1.082 to 1.809, while 7 inflammatory proteins (CCL25, CCL4, CSF-1, FGF-19, IL-10, IL-10RA, and IL-12B) were negatively associated with the risk of sporadic AD with OR value from 0.952 to 0.611, which suggesting their protective role in sporadic AD (Figure 6). Supplemental Figure 2 is the forest plot, scatter plot, funnel plot and leave-one-out analysis of identified inflammatory proteins IVs with causal effect on sporadic AD. MR Egger and Cochran's Q test suggest no heterogeneity and horizontal pleiotropy (Supplemental Table 5). Then these 17 inflammatory proteins were selected for meta-analysis using all sporadic AD datasets, respectively. The meta-analysis validated 2 inflammatory proteins (AXIN1 and IL-1 alpha) were positively associated with the risk of sporadic AD, with the OR = 1.08, 95% IC (1.02–1.14) and OR = 1.09, 95% IC (1.03–1.15), respectively (Figure 7). The detailed MR results of inflammatory proteins in sporadic AD were list in the Supplemental Table 6.

Figure 6.

The causal effect of inflammatory proteins on sporadic AD. Causal effect of 17 inflammatory proteins on sporadic AD in different outcome datasets. nsnp: number of single nucleotide polymorphisms; OR: odds ratio; CI: confidence interval; FDR: false discovery rate.

Figure 7.

Meta-analysis result of MR results between inflammatory proteins and sporadic AD. Above are forest charts of AXIN1 and IL 1 alpha, and below are the corresponding funnel charts to eliminate data bias.

Causal effect of 91 inflammatory factors on familial AD

For familial AD, LIF and AXIN1 showed playing weak risk roles with OR = 1.004 and 1.005 (p < 0.05) in outcome dataset ukb-b-14699, respectively. However, after FDR correction, no causal association was found between any of the 91 inflammatory proteins and 8 familial AD datasets (Supplemental Table 7).

Discussion

The gram-negative bacteria and their LPS significantly contribute to the pathogenesis of AD. However, the causal relationship between serum LPS activity level and the risk of AD has not been established. LPS often promotes the occurrence and development of AD through chronic inflammation, which is also a known risk factor for AD. In this study, we systematically analyzed the causal effects of serum LPS activity level and 91 circulating inflammatory proteins on AD (in both 1, 260, 136 sporadic and 2, 838, 825 familial AD samples) using a two-sample MR method and meta-analysis. The results showed that serum LPS activity level was as a risk factor for early onset sporadic AD, and a potential protective factor for familial AD. Two inflammatory proteins AXIN1 and IL-1 alpha have risk causal effects on sporadic AD. For familial AD, no inflammatory proteins were found to be associated with AD after meta-analysis. To our knowledge, this is the first study to investigate the causal effects between LPS and AD and to identify the opposite effects of serum LPS activity levels on sporadic and familial AD. These findings may provide a new perspective for the early diagnosis and treatment of AD.

LPS in the gut can translocate into the systemic circulation via dysfunction of the intestinal barrier, causing low-grade endotoxaemia.²⁴ Elevated levels of LPS in the blood and brain may promote microglial activation, amyloid and tau pathology changes, those were the pathological mechanism of AD.²⁵ Delivering a small amount of LPS to healthy volunteers can establish a human LPS inflammatory response model to study early human inflammation.²⁶ However, unexpectedly, in our study, serum LPS activity level only showed weak risk effect on early onset sporadic AD. The LPS-related GWAS dataset has been used to prove the association between genetic loci of serum lipopolysaccharide activity and contact activation pathway, vasoactivity, and lipoprotein metabolism, which have causal effect on thromboembolism and stroke.²⁷ That is to say the LPS-related GWAS dataset we have selected can be used for causal risk analysis of diseases. And, in our study, in familial AD patients, the same analysis was performed and the result showed that there is a potential negative correlation of the serum LPS activity level with the risk of AD. In the meta-analysis, the OR values of serum LPS activity level in multiple familial AD datasets are skewed towards the left, although the final result was not statistically significant (Figure 5). The opposite effect of serum LPS activity level on sporadic and familial AD is an important finding of this study. We speculated the low concentration serum LPS may not be sufficient to cause the occurrence of AD, both in sporadic and familial AD patients. Mizobuchi et al. reviewed that high-dose lipopolysaccharide is an inflammatory inducer while low-dose lipopolysaccharide is an immunomodulator, and their study demonstrated oral lipopolysaccharide can reduce AD pathology.²⁸ So, in our research results, serum LPS activity level was found to be a weak potential risk factor against sporadic AD. However, familial AD has clear heritability, mainly caused by mutations in APP, PSEN1, and PSEN2 genes. Maybe, serum LPS activity may exert immunomodulatory effects against hereditary pathogenesis, potentially conferring a protective role in the onset of familial AD.

The role of neuroinflammation in AD progression is increasingly supported by evidence and inflammation signaling pathways has become a potential target for the prevention of this neurodegenerative condition.^29–31 In this study, through the combination of GWAS and meta-analysis in a large population, we ultimately identified two circulating inflammatory factors closely associated with the risk of sporadic AD, AXIN1 and IL-1 alpha. It has been clarified that protein encoded by AXIN1 can suppress Wnt signaling and regulates the stabilization of beta-catenin.^32,33 While Wnt-β-catenin signaling serves as a protective mechanism against AD, and a prolonged disruption of Wnt signaling function could contribute to the Aβ-related neurodegeneration.^34,35 Restoring or activating Wnt/β-catenin pathway may be a potential treatment strategy for AD.^36,37 Our research demonstrated a direct causal relationship between AXIN1 and AD, which provided significant scientific value for treatment of sporadic AD based on the Wnt signaling. Additionally, in the recent MR research, AXIN1 was proved as a risk factor for AD, which consistent with our findings despite differences in research strategies and datasets.³⁸ In familial AD, AXIN1 also showed weak risk effect before FDR correction. The association between AXIN1 and familial AD is worth exploring in larger samples in the future.

Interleukin-1 alpha (IL-1 alpha) is another inflammatory protein associated with the risk of sporadic AD in our study. Du et al. reported the association between IL-1 alpha (−889) allele 2 polymorphism and later-onset AD with OR = 7.2 in homozygotes.³⁹ Murphy et al. found that IL-1 alpha (−889) 2 allele demonstrated accelerate the rate of cognitive decline in AD.⁴⁰ The homozygous IL-1 alpha T/T genotype was also showed association with AD onset age.⁴¹ Of course, there have also been some negative results reported in the study of the association between IL-1 and AD, considering it to be due to the different study populations.⁴² A meta-analysis data from 32 case-control studies suggested the association between IL-1 alpha −889C/T polymorphism and AD, though subgroup analysis revealed the robust associations in Caucasian populations but not in Asians.⁴³ A standard and cumulative meta-analysis found rs1800587 of IL-1 alpha polymorphism was significantly associated with higher AD risk in Caucasian compared with Asians. The observed ethnic differences may reflect low statistical power in Asian subgroups. Maybe future large-scale multi-ethnic studies using standardized methods could confirm these findings and uncover the association between genetic variations of IL-1 alpha and AD across populations.

While this study provides novel insights into the association between serum LPS activity or inflammatory proteins and AD risk, three key limitations warrant consideration. First, in the selection of genetic IVs in exposure factor serum LPS activity, due to too few IVs after removing the LD SNPs, the filtering criteria are only constrained by p-values and F value. SNPs associated with the serum LPS activity level with p-value <1 × 10⁻⁵ was only 71. Despite all this, only weak risk effect of the serum LPS activity level on early onset sporadic AD and potential protective effect on familial AD were observed. Second, since the existing dataset limitation, the study population is only limited to the European population and restricts generalizability, studies in multi-ethnic populations are recommended. Third, the meta-analysis relied exclusively on IVW method results, potentially introducing residual bias even after adjustments for heterogeneity and pleiotropy. Despite these limitations, this represents the most comprehensive MR study to date integrating meta-analytic approaches to investigate serum LPS and inflammatory proteins as risk factors for AD.

Conclusions

In summary, evidence from this large-scale cohort MR combined with meta-analysis study suggested that serum LPS activity level may present an opposite effect in sporadic and familial AD: protective effect on sporadic AD and potential risk effect on familial AD. Only two inflammatory proteins AXIN1 and IL-1 alpha were observed as risk factors in sporadic AD. These findings provide a new perspective for the early diagnosis and treatment of sporadic and familial AD. Further studies in multi-ethnic populations are recommended to confirm our conclusions.

Supplemental Material

sj-xlsx-1-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-1-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-2-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-2-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-3-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-3-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-4-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-4-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-5-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-5-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-6-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-6-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-xlsx-7-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-xlsx-7-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-docx-8-alr-10.1177_25424823251385589 - Supplemental material for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort

Supplemental material, sj-docx-8-alr-10.1177_25424823251385589 for Causal effect of serum lipopolysaccharide activity levels and inflammatory proteins on Alzheimer's disease: A Mendelian randomization study combined with meta-analysis in a large-scale cohort by Li Niu, Yubo Li, Hao Wu, Liping Zhao, Jin Zhang, Fan Lu, Guoqing Zhao, Fengfeng Jia, Jianjun Zhu and Ming Liu in Journal of Alzheimer's Disease Reports

Footnotes

Acknowledgements

We sincerely thank the researchers and participants of the original GWAS studies for their contributions. We also acknowledge the data platforms and consortiums for their efforts in collecting, curating, and sharing these valuable data.

ORCID iDs

Li Niu

Yubo Li

Hao Wu

Guoqing Zhao

Jianjun Zhu

Ming Liu

Ethical considerations

The publicly available databases including the Finngen database and IEU open GWAS database were used in this study and ethics approval and consent to participate were obtained in these studies.

Consent to participate

Not applicable

Consent for publication

Not applicable

Author contribution(s)

Li Niu: Conceptualization; Formal analysis; Writing – original draft; Writing – review & editing.

Yubo Li: Data curation; Methodology; Writing – original draft; Writing – review & editing.

Hao Wu: Data curation; Resources; Writing – review & editing.

Liping Zhao: Validation; Writing – review & editing.

Jin Zhang: Validation; Writing – review & editing.

Fan Lu: Writing – review & editing.

Guoqing Zhao: Validation; Writing – review & editing.

Fengfeng Jia: Writing – review & editing.

Jianjun Zhu: Conceptualization; Validation; Writing – original draft; Writing – review & editing.

Ming Liu: Conceptualization; Project administration; Supervision; Writing – original draft.

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 National Natural Science Foundation of China (No.81902513), Applied Basic Research Project of Shanxi Province (No. 20210302123319, 20210302124376 and 202103021224228), and Science Research Start-up fund for doctor of Shanxi Medical University (XD1808, BS03201603).

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 from the Finngen database (https://www.finngen.fi/en/access_results) and the publicly available IEU open GWAS database () were downloaded to analysis. The original analysis data can be checked in Supplementary Material of the article.

Supplemental material

Supplemental material for this article is available online.

References

Liu

Zhang

Wang

. Updates in Alzheimer's disease: from basic research to diagnosis and therapies. Transl Neurodegener 2024; 13: 45.

Scheltens

De Strooper

Kivipelto

, et al. Alzheimer's disease. Lancet 2021; 397: 1577–1590.

Rajan

Weuve

Barnes

, et al. Population estimate of people with clinical Alzheimer's disease and mild cognitive impairment in the United States (2020–2060). Alzheimers Dement 2021; 17: 1966–1975.

Alzheimer's Association . 2023 Alzheimer's disease facts and figures. Alzheimers Dement 2023; 19: 1598–1695.

Krishnamurthy

Jayaraman

Krishna

, et al. An overview of the genes and biomarkers in Alzheimer's disease. Ageing Res Rev 2024; 104: 102599.

Kim

Cho

, et al. APOE epsilon4-dependent effects on the early amyloid pathology in induced neurons of patients with Alzheimer's disease. Transl Neurodegener 2022; 11: 45.

Shi

Yamada

Liddelow

, et al. Apoe4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature 2017; 549: 523–527.

Gordon

Blazey

, et al. Spatial patterns of neuroimaging biomarker change in individuals from families with autosomal dominant Alzheimer's disease: a longitudinal study. Lancet Neurol 2018; 17: 241–250.

Gunasekaran

Reyes-Dumeyer

Faber

, et al. Missense and loss-of-function variants at GWAS loci in familial Alzheimer's disease. Alzheimers Dement 2024; 20: 7580–7594.

10.

Bertram

Lange

Mullin

, et al. Genome-wide association analysis reveals putative Alzheimer's disease susceptibility loci in addition to APOE. Am J Hum Genet 2008; 83: 623–632.

11.

Seshadri

Fitzpatrick

Ikram

, et al. Genome-wide analysis of genetic loci associated with Alzheimer disease. JAMA 2010; 303: 1832–1840.

12.

van Es

van den Berg

. Alzheimer's disease beyond APOE. Nat Genet 2009; 41: 1047–1048.

13.

Ganz

Fainstein

Ben-Hur

. When the infectious environment meets the AD brain. Mol Neurodegener 2022; 17: 53.

14.

Dominy

Lynch

Ermini

, et al. Porphyromonas gingivalis in Alzheimer's disease brains: evidence for disease causation and treatment with small-molecule inhibitors. Sci Adv 2019; 5: eaau3333.

15.

Kwon

Koh

. Neuroinflammation in neurodegenerative disorders: the roles of microglia and astrocytes. Transl Neurodegener 2020; 9: 42.

16.

Guo

Wang

Yin

. Microglia polarization from M1 to M2 in neurodegenerative diseases. Front Aging Neurosci 2022; 14: 815347.

17.

Singh

. Astrocytic and microglial cells as the modulators of neuroinflammation in Alzheimer's disease. J Neuroinflammation 2022; 19: 206.

18.

Sanchez-Tapia

Mimenza-Alvarado

Granados-Dominguez

, et al. The gut microbiota-brain axis during aging, mild cognitive impairment and dementia: role of tau protein, beta-amyloid and LPS in serum and curli protein in stool. Nutrients 2023; 15: 932.

19.

Andreadou

Katsipis

Tsolaki

, et al. Involvement and relationship of bacterial lipopolysaccharides and cyclooxygenases levels in Alzheimer's disease and mild cognitive impairment patients. J Neuroimmunol 2021; 357: 577561.

20.

Ohm

Hosseini

Lonnemann

, et al. The potential therapeutic role of itaconate and mesaconate on the detrimental effects of LPS-induced neuroinflammation in the brain. J Neuroinflammation 2024; 21: 207.

21.

Lista

Imbimbo

Grasso

, et al.

Tracking neuroinflammatory biomarkers in Alzheimer's disease: a strategy for individualized therapeutic approaches?

J Neuroinflammation 2024; 21: 187.

22.

Abdul Manap

Almadodi

Sultana

, et al. Alzheimer's disease: a review on the current trends of the effective diagnosis and therapeutics. Front Aging Neurosci 2024; 16: 1429211.

23.

Ference

Holmes

Smith

. Using Mendelian randomization to improve the design of randomized trials. Cold Spring Harb Perspect Med 2021; 11: a040980.

24.

Violi

Cammisotto

Bartimoccia

, et al. Gut-derived low-grade endotoxaemia, atherothrombosis and cardiovascular disease. Nat Rev Cardiol 2023; 20: 24–37.

25.

Brown

Heneka

. The endotoxin hypothesis of Alzheimer's disease. Mol Neurodegener 2024; 19: 30.

26.

Brooks

Barr

Wiscombe

, et al. Human lipopolysaccharide models provide mechanistic and therapeutic insights into systemic and pulmonary inflammation. Eur Respir J 2020; 56: 1901298.

27.

Leskela

Toppila

Harma

, et al. Genetic profile of endotoxemia reveals an association with thromboembolism and stroke. J Am Heart Assoc 2021; 10: e022482.

28.

Mizobuchi

Soma

. Low-dose lipopolysaccharide as an immune regulator for homeostasis maintenance in the central nervous system through transformation to neuroprotective microglia. Neural Regen Res 2021; 16: 1928–1934.

29.

Bradburn

Murgatroyd

Ray

. Neuroinflammation in mild cognitive impairment and Alzheimer's disease: a meta-analysis. Ageing Res Rev 2019; 50: 1–8.

30.

Almutary

Begum

Kyada

, et al. Inflammatory signaling pathways in Alzheimer's disease: mechanistic insights and possible therapeutic interventions. Ageing Res Rev 2025; 104: 102548.

31.

Botella Lucena

Heneka

. Inflammatory aspects of Alzheimer's disease. Acta Neuropathol 2024; 148: 31.

32.

Zeng

Fagotto

Zhang

, et al. The mouse fused locus encodes Axin, an inhibitor of the Wnt signaling pathway that regulates embryonic axis formation. Cell 1997; 90: 181–192.

33.

Kishida

Yamamoto

Ikeda

, et al. Axin, a negative regulator of the wnt signaling pathway, directly interacts with adenomatous polyposis coli and regulates the stabilization of beta-catenin. J Biol Chem 1998; 273: 10823–10826.

34.

Shi

Shen

Shi

, et al. Osteocyte-derived sclerostin impairs cognitive function during ageing and Alzheimer's disease progression. Nat Metab 2024; 6: 531–549.

35.

Inestrosa

Toledo

. The role of Wnt signaling in neuronal dysfunction in Alzheimer's disease. Mol Neurodegener 2008; 3: 9.

36.

Wang

Huang

, et al. Activation of Wnt/beta-catenin pathway mitigates blood-brain barrier dysfunction in Alzheimer's disease. Brain 2022; 145: 4474–4488.

37.

Yoon

Kim

Shin

, et al. Inhibition of CXXC5 function rescues Alzheimer's disease phenotypes by restoring Wnt/beta-catenin signaling pathway. Pharmacol Res 2023; 194: 106836.

38.

Shang

Liu

Gong

, et al. Exploring the bidirectional relationships between Alzheimer's disease and cerebral small vessel disease: insights from Mendelian randomization. J Stroke Cerebrovasc Dis 2025; 34: 108259.

39.

Dodel

Eastwood

, et al. Association of an interleukin 1 alpha polymorphism with Alzheimer's disease. Neurology 2000; 55: 480–483.

40.

Murphy

Jr. Claassen

DeVoss

, et al. Rate of cognitive decline in AD is accelerated by the interleukin-1 alpha −889 *1 allele. Neurology 2001; 56: 1595–1597.

41.

Grimaldi

Casadei

Ferri

, et al. Association of early-onset Alzheimer's disease with an interleukin-1alpha gene polymorphism. Ann Neurol 2000; 47: 361–365.

42.

Kim

, et al. Lack of association of the interleukin-1alpha gene polymorphism with Alzheimer's disease in a Korean population. Ann Neurol 2001; 49: 817–818.

43.

Qin

Peng

Zeng

, et al. Interleukin-1A −889C/T polymorphism and risk of Alzheimer's disease: a meta-analysis based on 32 case-control studies. J Neurol 2012; 259: 1519–1529.

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.

0.01 MB

0.02 MB

0.00 MB

0.03 MB

10.56 MB

0.34 MB

0.47 MB

0.01 MB

0.02 MB