Causal Effects of Kidney Function and Chronic Kidney Disease on Alzheimer’s Disease by Analyzing Large-Scale Genome-Wide Association Study Datasets

Abstract

Background

Alzheimer’s disease (AD) is the leading cause of dementia. Genetic components play an important role in AD and have been widely evaluated by genome-wide association studies (GWAS) and exome sequencing, and some common and rare genetic variants have been identified. In addition to genetic factors, environment factors have a role in AD. Growing evidence from observational studies linked impaired kidney function to cognitive impairment and AD; however, there are inconsistences in these findings.

Objective

To determine the causal effects of impaired kidney function and chronic kidney disease (CKD) on AD.

Methods

Mendelian randomization (MR) methods have been widely used to infer causal associations between exposure and outcome. Here, we conducted an MR study to investigate the causal effects of impaired kidney function and CKD on the risk of AD by analyzing large-scale GWAS datasets from FinnGen and CKD Genetics (CKDGen) Consortium.

Results

We found no significant but a suggestive effect of CKD on decreased risk of AD using inverse-variance weighted (IVW) (p = 8.46E–02) and simple mode (p = 7.60E–02) methods. We identified a statistically significant effect of the estimated glomerular filtration rate (eGFR) on increased risk of AD using IVW (p = 1.11E–02), weighted median regression (p = 5.60E–03), and weighted mode (p = 2.45E–02) methods.

Conclusions

Together, our findings indicate that high eGFR levels may increase the risk of AD. These findings need to be verified in future studies.

Keywords

Alzheimer’s disease chronic kidney disease genome-wide association study kidney function Mendelian randomization

Introduction

Alzheimer’s disease (AD) is the leading cause of dementia.^1–3 Genetic components play an important role in AD, and have been widely evaluated by genome-wide association studies (GWAS) and exome sequencing, and some common and rare genetic variants have been identified.^{1,4, 1,4} In addition to genetic factors, environment factors are involved in AD. Growing evidence from observational studies linked impaired kidney function to cognitive impairment and AD; however, there are inconsistences in these findings.^5–9

An observational study from the Rush Memory and Aging Project showed that a low estimated glomerular filtration rate (eGFR) at baseline was associated with rapid cognitive decline in 886 elderly people without dementia.⁵ Sixteen years of follow-up of 191,970 elderly participants without dementia in the UK Biobank showed that moderately to severely impaired kidney function increased the risk of all-cause dementia (hazard ratio (HR) = 1.53, 95% CI: 1.32–1.76), vascular dementia (HR = 1.51, 95% CI: 1.13–2.02), and AD (HR = 1.27, 95% CI: 1.00–1.60, p = 0.046).⁷

Kjaergaard and colleagues conducted a meta-analysis of 468,699 Scandinavians⁹ and found that compared with the reference eGFR, mildly impaired (HR = 1.14, 95% CI: 1.06–1.22) and severely impaired (HR = 1.91, 95% CI: 1.21–3.01), but not moderately impaired (HR = 1.31, 95% CI: 0.92–1.87) eGFR significantly increased the risk of all-cause dementia.⁹ However, only impaired (HR = 1.24, 95% CI: 1.08–1.43), but not moderately impaired (HR = 1.32, 95% CI: 0.87–1.99) or severely impaired (HR = 1.19, 95% CI: 0.70–2.04) eGFR significantly increased the risk of AD.⁹ A recent 17 years of follow-up of 6,256 participants indicated that impaired kidney function did not increase the risk of all-cause dementia, AD, or vascular dementia.⁸ Xu and colleagues evaluated the relation between eGFR and the risk of dementia in 329,822 elderly residents of Stockholm without dementia or kidney replacement therapy.⁶ They found that compared with elderly residents with eGFR levels of 90–104 mL/min, elderly residents with eGFR levels≥105 mL/min had increased risk of all-cause dementia (HR = 1.30, 95% CI: 0.90–1.87, p = 0.16), vascular dementia (HR = 2.00, 95% CI: 1.02–3.95, p = 0.04), but not AD (HR = 0.76, 95% CI: 0.37–1.59, p = 0.47).⁶

These inconsistent findings of the relationship between impaired kidney function and AD suggest that observational evidence may be problematic because observational studies often include confounding factors and reverse causality.¹⁰ Mendelian randomization (MR) is an alternative and widely used way to infer causal associations between exposure and outcome.¹⁰ Here, we conducted a MR study to investigate the causal effects of impaired kidney function and chronic kidney disease (CKD) on the risk of AD by analyzing large-scale GWAS datasets.

Materials and methods

Study design

We performed two-sample MR to evaluate the causal effects of impaired kidney function and CKD on the risk of AD using large-scale publicly available and independent kidney function, CKD, and AD GWAS summary statistics. The study design is shown in Figure 1.

Figure 1.

Study design used in this study.

eGFR, CKD, and AD GWAS datasets

The eGFR and CKD GWAS datasets were obtained from the CKD Genetics (CKDGen) Consortium.¹¹ The CKDGen Consortium performed a large-scale trans-ancestry eGFR GWAS meta-analysis of 765,348 individuals of European, East Asian, African-American, South Asian, and Hispanic ancestry and identified 308 eGFR genetic variants with genome-wide significance.¹¹ The CKDGen Consortium also conducted a large-scale trans-ancestry CKD GWAS meta-analysis of 625,219 individuals, including 64,164 CKD cases, and identified 23 CKD genetic variants with genome-wide significance.¹¹ The AD GWAS dataset was from FinnGen R10 and consisted of 2,419 AD cases and 183,753 controls; AD was defined to be atypical or mixed with more control exclusions (AD_AM_EXMORE).¹² Here, we selected the 308 eGFR genetic variants and 23 CKD genetic variants identified by the CKDGen Consortium GWAS meta-analysis as the potential instrumental variables (IVs). A summary of the GWAS summary data used in the present study is provided in Table 1, and detailed information about the 2,419 AD cases from FinnGen R10 is provided in Figure 2.

Table 1.

Summary of the GWAS data used this study.

Source	Phenotype	Abbr.	Cases	Controls	All
CKDGen	Estimated glomerular filtration rate	eGFR	–	–	765,348
CKDGen	Chronic kidney disease	CKD	64,164	561,055	625,219
FinnGen	Alzheimer’s disease	AD	2,419	183,753	186,172

GWAS, genome-wide association study; CKDGen, Chronic Kidney Disease Genetics Consortium; FinnGen, FinnGen R10.

Figure 2.

Detailed information of the 2,419 Alzheimer’s disease cases from FinnGen R10.

Mendelian randomization analysis

MR analysis was performed using five methods, inverse-variance weighted meta-analysis (IVW), weighted median regression, MR-Egger, simple mode, and weighted mode, that are well established and widely described.13 –17 We selected IVW as the main analysis method and the others as the sensitivity analysis methods. Potential pleiotropy was evaluated using both the heterogeneity test from Cochrane’s Q statistic and pleiotropy test from MR-Egger and Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO).18,19, 18,19 All the statistical analyses were conducted using R software (v4.2.2), and the meta, MendelianRandomization,²⁰ and TwoSampleMR¹⁵ R packages. The causal effects of eGFR and CKD were reflected using the odds ratio (OR) and 95% confidence interval (CI), which correspond to the AD risk for each standard deviation increase in eGFR levels and CKD. The statistically significant causal association was defined as p < 0.05.

Strength of instrumental variables

We calculate the proportion of eGFR variance explained by the IVs using R² as:

\begin{aligned} R^{2} = \sum_{i = 1}^{k} 2 & * E A F & * (1 - E A F) & * β_{i}^{2} / variance \end{aligned},

where β_i is the effect size (beta) for IV_i, se(β_i) is the standard error for IV_i, k is the number of IVs,²¹ and variance is the variance of the sex- and age-adjusted log(eGFR) residuals. We assumed variance to be 0.016, which was calculated using the data from 11,827 European-ancestry participants in the Atherosclerosis Risk in Communities (ARIC) study, as provided in the original study.¹¹

We calculate the proportion of CKD variance explained by the IVs using R² as:

\begin{aligned} R^{2} = \sum_{i = 1}^{k} 2 & * (1 - E A F) & * β_{i}^{2}, \end{aligned}

where β_i is the effect size (beta) for IV_i, se(β_i) is the standard error for IV_i, and k is the number of IVs.²¹

Using R², we evaluated the strength of IVs by calculating the F statistic as:

F = \frac{R^{2} (N - 1 - k)}{k (1 - R^{2})},

where R² is the proportion of eGFR or CKD variance explained by the IVs, N is the sample size from the eGFR or CKD GWAS, and k is the number of IVs. IVs with F > 10 were considered to be robust IVs.²² A leave-one-out analysis was also conducted by leaving each IV out of the MR analysis in turn to assess whether the overall causal estimate was driven by one IV.²³

Results

Strength of instrumental variables

The MR analyses were conducted using 308 eGFR and 23 CKD genetic variants as the potential IVs; 304 of the eGFR genetic variants and the 23 CKD genetic variants were available in AD GWAS summary datasets provided in Supplementary Tables 1 and 2. After excluding palindromic genetic variants (A/T or C/G), 252 eGFR and 18 CKD genetic variants are selected for the MR analyses.

The proportion of eGFR variance explained by the 252 eGFR genetic variants was 6.86%, and the proportion of CKD variance explained by the 18 CKD genetic variants was 5.61%. All 252 eGFR and 18 CKD genetic variants as the IVs had F > 10; indeed, all 18 CKD genetic variants had minimum F > 644 and all 252 eGFR genetic variants had minimum F > 62 (Supplementary Tables 3 and 4). The leave-one-out analysis did not identify a single IV that affected the overall causal estimate (Supplementary Tables 3 and 4).

Mendelian randomization analysis

No statistically significant causal association was detected between CKD and AD using IVW, weighted median regression, MR-Egger, simple mode, or weighted mode (Table 2); however, a suggestive association was detected between CKD and decreased risk of AD using IVW (p = 8.46E–02) and simple mode (p = 7.60E–02) (Table 2).

Table 2.

Causal effects of eGFR and CKD on the risk of AD.

Exposure	Method	Beta	SE	95% CI	p
CKD	MR-Egger	0.144	0.272	–0.388, 0.677	6.03E–01
	Weighted median	0.000	0.149	–0.291, 0.291	9.99E–01
	Inverse variance weighted	–0.194	0.112	–0.414, 0.026	8.46E–02
	Simple mode	–0.570	0.302	–1.161, 0.021	7.60E–02
	Weighted mode	–0.048	0.169	–0.379, 0.284	7.82E–01
eGFR	MR-Egger	3.685	1.905	–0.050, 7.419	5.43E–02
	Weighted median	3.383	1.221	0.989, 5.777	5.60E–03
	IVW	1.945	0.766	0.443, 3.446	1.11E–02
	Simple mode	4.077	3.301	–2.393, 10.547	2.18E–01
	Weighted mode	4.351	1.923	0.582, 8.119	2.45E–02

eGFR, estimated glomerular filtration rate; CKD, chronic kidney disease; AD, Alzheimer’s disease; SE, standard error; CI, confidence interval; MR, Mendelian randomization; IVW, inverse-variance weighted. A statistically significant causal association was defined as p < 0.05.

We found a statistically significant causal association between eGFR and AD using IVW (p = 1.11E–02), weighted median regression (p = 5.60E–03), and weighted mode (p = 2.45E–02) (Table 2). Each standard deviation increase in eGFR levels was significantly associated with increased risk of AD (beta = 1.945, standard error = 0.766, 95% CI: 0.443–3.446, p = 1.11E–02). Importantly, the directions of MR estimates from MR-Egger and simple mode were consistent with those from IVW, weighted median regression, and weighted mode; however, no statistically significant causal association was detected using MR-Egger and simple mode (Table 2). The forest plot for overall causal effects of eGFR and CKD on the risk of AD using five MR methods is shown in Figure 3. The scatter plot of the individual causal effects of eGFR and CKD on the risk of AD using five MR methods is shown in Figure 4.

Figure 3.

Forest plot of the overall causal effects of eGFR and CKD on the risk of AD using five MR methods. eGFR, estimated glomerular filtration rate; CKD, chronic liver disease; AD, Alzheimer’s disease; MR, Mendelian randomization.

Figure 4.

Scatter plot of the individual causal effects of eGFR and CKD on the risk of AD using five MR methods. eGFR, estimated glomerular filtration rate; CKD, chronic liver disease; AD, Alzheimer’s disease; MR, Mendelian randomization.

Pleiotropy analysis

No clear pleiotropy was detected using the heterogeneity test with Cochrane’s Q statistic p > 0.05 in IVW and MR-Egger. The MR-Egger intercept test and MR-PRESSO global test also supported the lack of clear pleiotropy (Table 3). Together, all these findings show that the 252 eGFR and 18 CKD genetic variants met the IV assumptions.

Table 3.

Heterogeneity test and pleiotropy test of MR results.

	MR-Egger		IVW		MR-Egger			MR-PRESSO
Exposure	Q	p	Q	p	intercept	SE	p	Global Test p
CKD	18.7559	0.2815	20.9250	0.2297	–0.0291	0.0214	0.1926	0.2130
eGFR	259.4868	0.3268	260.5191	0.3266	–0.0058	0.0058	0.3196	0.1487

MR, Mendelian randomization; IVW, inverse-variance weighted; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate.

Discussion

Observational studies have produced inconsistent findings about the association between impaired kidney function and AD.5 –9 Here, we performed an MR study to evaluate the causal effects of eGFR and CKD on the risk of AD by analyzing large-scale GWAS datasets. We found no significant but a suggestive effect of CKD on decreased risk of AD using IVW (p = 8.46E–02) and simple mode (p = 7.60E–02). We identified a statistically significant effect of eGFR on increased risk of AD using IVW (p = 1.11E–02), weighted median regression (p = 5.60E–03), and weighted mode (p = 2.45E–02).

These findings show that CKD may reduce the risk of AD, and genetically high eGFR levels may increase the risk of AD. Our MR results are consistent with those of recent observational studies. Kjaergaard and colleagues conducted a Danish nationwide cohort study that included 82,690 patients with kidney disease and 413,405 healthy controls from the general population.²⁴ They found that kidney disease was significantly associated with increased risk of all-cause dementia (HR = 1.08, 95% CI: 1.03–1.12) and vascular dementia (HR = 1.26, 95% CI: 1.14–1.40), but associated with decreased risk of AD (HR = 0.85, 95% CI: 0.78–0.92).²⁴ A cohort study of 4,320,824 Denmark individuals > 40 years old showed that kidney disease, including CKD, was associated with the risk of all-cause dementia (HR = 1.15, 95% CI: 1.10–1.20) and vascular dementia (HR = 1.43, 95% CI: 1.33–1.54), but not associated with the risk of AD (HR = 0.98, 95% CI: 0.94–1.03).²⁵

Our MR findings are the first to highlight the causal association between kidney function and AD. Two previous MR studies evaluated the causal effect of kidney function on the risk of AD.9,26, 9,26 Liu and colleagues used three kidney function measures, eGFR, CKD, and blood urea nitrogen (BUN), and evaluated their effects on six neurodegenerative diseases, namely AD, multiple sclerosis, Parkinson’s disease, frontotemporal dementia, amyotrophic lateral sclerosis, and Lewy body dementia.²⁶ They selected 264 independent BUN, 147 independent CKD, and 308 independent eGFR genetic variants.²⁶ However, only 93 BUN, 27 CKD, and 191 eGFR genetic variants were available in the AD GWAS dataset from a meta-analysis of the International Genomics of Alzheimer’s Project GWAS stage 1 (21,982 AD and 41,944 controls) and UK Biobank (53,042 proxy AD and 355,900 controls) datasets.²⁷ The MR analysis showed lack of statistically significant effects of eGFR, CKD, or BUN on the risk of AD.²⁶ Kjaergaard and colleagues selected four kidney function measures, eGFR, BUN, CKD, and the urinary albumin-creatinine ratio (UACR), and evaluated their effects on AD using multiple AD GWAS datasets such as FinnGen R5, IGAP, IGAP/UKB and ADSP/IGAP/PGC-ALZ/UKB.⁹ They selected 169 eGFR, 24 CKD, 24 BUN, and 55 UACR genetic variants as the potential IVs; however, no statistically significant association of impaired kidney function with the risk of all-cause dementia and AD was detected.⁹

In summary, our findings indicate that CKD was not significantly associated with the risk of AD, whereas high eGFR levels may increase the risk of AD. These findings need to be verified in future studies using large-scale sample sizes.

Acknowledgments

We thank Margaret Biswas, PhD, from Liwen Bianji (Edanz) (http://www.liwenbianji.cn/) for editing the English text of a draft of this manuscript.

Supplemental Material

Supplementary Table 1

Supplementary Table 2

Supplementary Table 3

Supplementary Table 4

Footnotes

Author contributions

Hainan Zhao (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing); Hongxia Yuan (Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software); Ermin Wang (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration).

Funding

This work was supported by funding from the Project Guided by the Jinzhou Science and Technology Bureau (JZ2023B091), and the General Project of Liaoning Provincial Department of Science and Technology (2021-MS-335).

Declaration of conflicting interests

The authors have no conflict of interest to report.

Data availability

All relevant data are within the paper. The authors confirm that all data underlying the findings are either fully available without restriction through consortia websites, or may be made available from consortia upon request.

Supplemental material

The supplementary material is available in the electronic version of this article: .

References

Holstege

Hulsman

Charbonnier

, et al. Exome sequencing identifies rare damaging variants in ATP8B4 and ABCA1 as risk factors for Alzheimer’s disease. Nat Genet 2022; 54: 1786–1794.

Klomparens

and Ding

. Updates on the association of brain injury and Alzheimer’s disease. Brain Circ 2020; 6: 65–69.

Mangal

and Ding

. Mini review: Prospective therapeutic targets of Alzheimer’s disease. Brain Circ 2022; 8: 1–5.

Bellenguez

Kucukali

Jansen

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

Buchman

Tanne

Boyle

, et al. Kidney function is associated with the rate of cognitive decline in the elderly. Neurology 2009; 73: 920–927.

Garcia-Ptacek

Trevisan

, et al. Kidney function, kidney function decline, and the risk of dementia in older adults: A registry-based study. Neurology 2021; 96: e2956–e2965.

Wang

Guo

, et al. Association of kidney function with dementia and structural brain differences: A large population-based cohort study. J Gerontol A Biol Sci Med Sci 2024; 79: glad192.

Stocker

Beyer

Trares

, et al. Association of kidney function with development of Alzheimer disease and other dementias and dementia-related blood biomarkers. JAMA Netw Open 2023; 6: e2252387.

Kjaergaard

Ellervik

Witte

, et al. Kidney function and risk of dementia: Observational study, meta-analysis, and two-sample mendelian randomization study. Eur J Epidemiol 2022; 37: 1273–1284.

10.

Skrivankova

Richmond

Woolf

BAR

, et al. Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian randomization: The STROBE-MR Statement. JAMA 2021; 326: 1614–1621.

11.

Wuttke

, et al. A catalog of genetic loci associated with kidney function from analyses of a million individuals. Nat Genet 2019; 51: 957–972.

12.

Kurki

Karjalainen

Palta

, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023; 613: 508–518.

13.

Bowden

Davey Smith

Haycock

, et al. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 2016; 40: 304–314.

14.

Burgess

and Thompson

. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol 2017; 32: 377–389.

15.

Hemani

Zheng

Elsworth

, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife 2018; 7: e34408.

16.

Burgess

Foley

Allara

, et al. A robust and efficient method for Mendelian randomization with hundreds of genetic variants. Nat Commun 2020; 11: 376.

17.

Bowden

Davey Smith

and Burgess

. Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. Int J Epidemiol 2015; 44: 512–525.

18.

Davies

Holmes

and Davey Smith

. Reading Mendelian randomisation studies: A guide, glossary, and checklist for clinicians. BMJ 2018; 362: k601.

19.

Verbanck

Chen

Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

20.

Yavorska

and Burgess

. MendelianRandomization: An R package for performing Mendelian randomization analyses using summarized data. Int J Epidemiol 2017; 46: 1734–1739.

21.

Shim

Chasman

Smith

, et al. A multivariate genome-wide association analysis of 10 LDL subfractions, and their response to statin treatment, in 1868 Caucasians. PLoS One 2015; 10: e0120758.

22.

Burgess

and Thompson

. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 2011; 40: 755–764.

23.

Burgess

Davey Smith

Davies

, et al. Guidelines for performing Mendelian randomization investigations. Wellcome Open Res 2019; 4: 186.

24.

Kjaergaard

Johannesen

Sorensen

, et al. Kidney disease and risk of dementia: A Danish nationwide cohort study. BMJ Open 2021; 11: e052652.

25.

Saima

Mette Brimnes

Jan

, et al. Is kidney disease associated with both Alzheimer’s disease and vascular dementia?Alzheimers Dement 2020; 16: e042527.

26.

Liu

, et al. Renal function and neurodegenerative diseases: A two-sample Mendelian randomization study. Neurol Res 2023; 45: 456–464.

27.

Schwartzentruber

Cooper

Liu

, et al. Genome-wide meta-analysis, fine-mapping and integrative prioritization implicate new Alzheimer’s disease risk genes. Nat Genet 2021; 53: 392–402.