The Association of Serum Uric Acid Level,Gout,and Alzheimer’s Disease: A Bidirectional Mendelian Randomization Study

Abstract

Background:

The relationship between serum uric acid (UA) and Alzheimer’s disease (AD) risk still remained ambiguous despite extensive attempts.

Objective:

Via the two-sample Mendelian randomization (MR) design, we aimed to examine the bidirectional causal relationships of serum UA, gout, and the risk of AD.

Methods:

Genetic variants of UA, gout, and AD were extracted from published genome-wide association summary statistics. The inverse-variance weighted (IVW, the primary method), and several sensitivity methods (MR-Egger, weighted median, and weighted mode) were used to calculate the effect estimates. Egger regression, MR-PRESSO and leave-one-SNP-out analysis were performed to identify potential violations.

Results:

Genetic proxies for serum UA concentration [odds ratio (OR_IVW) = 1.09, 95% confidence interval (CI) = 1.01–1.19, p = 0.031] were related with an increased risk of AD using 25 single nucleotide polymorphisms (SNPs). This causal effect was confirmed by sensitivity analyses including MR-Egger (1.22, 1.06–1.42, p = 0.014), weighted median (1.18, 1.05–1.33, p = 0.006), and weighted mode (1.20, 1.07–1.35, p = 0.005) methods. No evidence of notable directional pleiotropy and heterogeneity were identified (p > 0.05). Three SNPs (rs2078267, rs2231142, and rs11722228) significantly drove the observed causal effects. Supportive causal effect of genetically determined gout on AD risk was demonstrated using two SNPs (OR_IVW = 1.05, 95% CI = 1.00–1.11, p = 0.057). No reverse causal effects of AD on serum UA levels and gout risk were found.

Conclusion:

The findings revealed a causal relationship between elevated serum UA level and AD risk. However, further research is still warranted to investigate whether serum UA could be a reliable biomarker and therapeutic target for AD.

Keywords

Alzheimer’s disease gout Mendelian randomization uric acid

INTRODUCTION

The incidence of dementia is increasing at an impressive rate due to increased lifespan, becoming one of the greatest global challenges for public health and social care [1]. Alzheimer’s disease (AD), the most common type of dementia, accounts for an estimated 60% to 80% of the dementia cases [2]. It is a complex neurodegenerative brain disorder, typically manifested by progressive loss of memory and cognitive function and considered to be a biological and clinical continuum [3, 4]. The pathophysiological mechanisms underlying AD remain unclear, but oxidative stress, the imbalance between free radical production, scavenging, and antioxidant defenses, has been suggested to be involved in several vital pathways in the pathogenesis of AD [5, 6].

Serum uric acid (UA), normally regarded as an important endogenous antioxidant, accounts for approximately 60% proportion of the free radical scavenging capacity [7], but it can also act as a pro-oxidant depending on its chemical status and microenvironment [8, 9]. Despite extensive studies exploring the correlation between UA and dementia risk, the conclusions remained ambiguous. On the one hand, evidence revealed that UA could protect or delay the pathological processes of neurodegeneration, mainly by scavenging free radicals and reducing oxidative stress response [10]. On the other hand, elevated serum UA levels are also linked to several chronic diseases, including hyperglycemia, diabetes mellitus, obesity, and hypertension, well-known risk factors for AD dementia [11]. Observational studies provided opposite conclusions, with one longitudinal cohort study revealing a protective effect of high serum UA concentration on dementia risk [12] and another presenting the harmful role [13]. Besides, the association between gout, a prevalent inflammatory arthritis mainly caused by hyperuricemia [14], and dementia has been explored in many large-scale cohorts, however, with conflicting results. Two previous cohort studies, one in the UK [15] and another in Taiwan [16], both reported lower AD incidence in gout patients. On the contrary, Singh et al. discovered an increased risk of dementia in patients with gout in a large sample of US participants [17]. Obviously, more research is needed to better understand the association between UA, gout, and dementia risk.

In spite of this, existing studies are still limited to observational study design, which is subject to unmeasured confounding bias and reverse causation. Randomized controlled trials (RCTs) design could help establish the causal relationship of serum UA and AD risk, but trial data are scarce. An alternative approach to determine causality is Mendelian randomization (MR) design. In MR analysis, genetic proxies such as single nucleotide polymorphisms (SNPs) are regarded as instrumental variables (IVs) to represent the lifetime changes of the trait of interest [18]. The MR approach has been widely employed to help uncover the causal effects of many risk factors on the development of AD [19 –23]. Over the last few years, several MR studies have attempted to illustrate the causality between UA, AD dementia, and cognition, resulting in inconsistent conclusions [24 –27]. Considering that the more recent and larger genome-wide association studies (GWASs) are available, we have more opportunities to study these causal relationships with convincing power and minimize bias as much as possible.

Here in this study, by using published summary-level data from large genetic GWASs, we designed a bidirectional two-sample MR analysis to determine whether genetically predicted serum UA level and gout are causally related to the risk of AD.

MATERIALS AND METHODS

Study design

We adopted a bidirectional two-sample MR design, which is a genetic IV analysis based on summary-level data with SNPs as instruments for the exposures of interest. IVs are selected to mimic the randomization of the risk factors, ensuring compatibility of groups with respect to any measured/unmeasured confounders. There are three critical assumptions that must hold: 1) the genetic variants, e.g., the SNPs, must be strongly associated with the exposures of interest; 2) the genetic instruments are independent of the unmeasured confounders; 3) the genetic instruments are only related to the outcome through the exposure of interest, not by the confounders [28, 29] (Fig. 1). In this bidirectional MR, serum UA/gout, and AD mutually served as exposure and outcome, respectively.

Fig. 1

Scheme diagram of Mendelian randomization design. This is a two-sample, bidirectional MR analysis. Serum uric acid/gout, and AD mutually serve as exposure and outcome, respectively. The MR approach should satisfy three assumptions: 1) the genetic variants significantly associate with the exposures; in this study, the two exposures are serum UA levels or gout, and the AD; 2) the IVs (namely, SNPs) have no association with confounding factors; and 3) the risk of outcome (AD, and serum UA or gout) is influenced only by the exposure, not by other pathways. AD, Alzheimer’s disease; IV, instrumental variable; SNP, single nucleotide polymorphism; UA, uric acid.

Serum uric acid and gout database

Genetic variants of serum UA concentration and gout were extracted from a GWAS meta-analysis that combined data from over 140,000 European individuals within the Global Urate Genetics Consortium (GUGC) [30]. A total of 110,347 individuals from 48 studies contributed to the GWAS meta-analysis of serum UA concentration and 69,374 participants (N = 2,115 cases, N = 67,259 controls) from 14 studies contributed to the gout’s meta-analysis. This GWAS meta-analysis is the largest GWAS of UA and gout so far. The mean age of included participants ranged from 25.5 (3.3) to 76.4 (5.5) years old. The mean serum UA concentrations ranged from 3.9 to 6.1 mg/dl and the proportion with gout ranged from 0.9% to 6.4%. Detailed information on study design, inclusion/exclusion processes, UA measurement and gout definition are provided previously [30].

Alzheimer’s disease database

Genetic variants of AD were extracted from a recent large GWAS meta-analysis of non-Hispanic Whites within the International Genomics of Alzheimer’s Project (IGAP) GWAS Stage 1 sample of the 46 datasets from four consortia (N = 21,982 cases, N = 41,944 controls) [31]. Detailed demographics of all 46 case-control studies from the four consortia are described previously. The mean age of disease onset ranged from 71.1 to 82.6 years in patients with AD, and the age ranged from 51.0 to 78.9 years in cognitively normal controls. AD cases were diagnosed by accepted diagnostic criteria, either autopsy- or clinically-confirmed. Further details of the cohorts and analytical process can be found in the original literature [31].

Statistical analyses

The summary statistics were combined to estimate the causal relations between serum UA levels, gout, and AD using four methods, including inverse variance weighting (IVW), MR-Egger, weighted median, and weighted mode, which provides different assumptions about horizontal pleiotropy [32]. The IVW method was considered as our main method, providing robust causal estimates in the absence of directional pleiotropy. It assumes the intercept is zero and calculates the weighted regression of SNP-exposure effects with SNP-outcome effects [33]. The causal estimate of MR Egger regression represents a genotype-outcome dose response relationship which takes pleiotropic effects into account, given that it is not constrained to have a slope through zero [32, 34]. Statistically significant intercept terms imply the evidence of unbalanced pleiotropy. The weighted median approach is defined as the median of a weighted empirical density function of the ratio estimates, which gives more weight to more accurate IVs. The estimates are consistent even if up to 50% of the information comes from invalid or weak IVs [29]. The weighted mode method requires that the most common estimates causal effects be consistent estimates of true causal effects, even if the majority of instruments are invalid [35]. All of the results were presented as odds ratios (ORs) with 95% confidence intervals (CIs).

Heterogeneity and directional pleiotropic effects was measured by the Egger intercept [36], Cochran Q-derived IVW estimate [37] and MR pleiotropy residual sum and outlier (MR-PRESSO) test [38]. The intercept term in Egger regression could clearly indicate whether directional horizontal pleiotropy drives the results of MR analyses. The Cochran Q statistic uses modified second order weights and takes into account the uncertainty of the numerator and denominator of the IV ratio, which alleviates the no-measurement-error assumption [37]. The MR-PRESSO method excludes outliers determined by the residual square errors from the regression of SNP-outcome against SNP-exposure to compute an estimate free from outlier effects [38]. In case of evidence of horizontal pleiotropy, the approach compares expected and observed distribution of individual variants to identify outliers. Moreover, “leave-one-out” validation analysis by systematic removal of genetic instruments from MR analysis, were performed to identify if a single SNP was driving the association. F statistics were used to measure the strength of genetic instruments in IVW [39]; F greater than 10, indicates that the instrument efficacy is sufficient, and the MR analysis is less likely to be influenced by weak instrument bias [40].

Statistical significance was set at a 2-sided multiple corrected p value < 0.025 (0.05/2, two variables, UA and gout). Analyses were performed in R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) by using the “TwoSampleMR” package.

RESULTS

Instrument identification

Using genome-wide significant (p < 5×10^–⁸) SNPs from large GWAS, we estimated the overall effect of serum UA and gout on the risk of AD. The selected significant SNPs were subsequently clumped to a default threshold of linkage disequilibrium (LD R² < 0.001) based on European 1000 Genomes. To eliminate the genetic bias caused by palindrome with intermediate allele frequencies, we further excluded several SNPs. Eventually, the screening process led to an inclusion of 25 SNPs for UA-AD and 2 SNPs for gout-AD association pairs. The SNPs were reported to be located in UA or gout related locus, e.g., SLC17A1, TRIM46, SLC16A9, SLC2A9, BAZ1B, etc. The detailed information of the included SNPs was shown in the Table 1. Next, we treated the AD risk as the exposure to investigate the effect of AD on serum UA levels and gout risk. In these analyses, 4 SNPs were included respectively (Supplementary Table 1). All of the SNPs had F statistics higher than 10.

Table 1

The information of included SNPs on AD effects

Traits	SNP	EA	Gene	Position	Beta (SE) for exposure	p for exposure
UA	rs10761587	T	NA	10:50889433	0.062 (0.009)	9.58E-11
UA	rs11264341	T	TRIM46	1:155179017	–0.048 (0.006)	1.04E-14
UA	rs1165151	T	SLC17A1	6:25821388	–0.092 (0.005)	4.52E-60
UA	rs1171614	T	SLC16A9	10:59709780	–0.074 (0.007)	6.48E-23
UA	rs11722228	T	SLC2A9	4:9914117	0.210 (0.006)	1.00E-200
UA	rs1178977	A	BAZ1B	7:73442719	0.050 (0.007)	6.68E-12
UA	rs1260326	T	GCKR	2:27508073	0.077 (0.006)	1.31E-40
UA	rs1394125	A	UBE2Q2	15:75866642	0.043 (0.006)	9.78E-11
UA	rs1471633	A	NA	1:145711327	0.061 (0.005)	1.40E-26
UA	rs17050272	A	LOC105373585	2:120548864	0.037 (0.006)	9.36E-09
UA	rs1825043	A	NA	4:9572116	0.100 (0.013)	6.67E-14
UA	rs2078267	T	SLC22A11	11:64566642	–0.078 (0.006)	8.73E-36
UA	rs2231142	T	ABCG2	4:88131171	0.220 (0.009)	4.43E-116
UA	rs2307394	T	ORC4	2:147958859	–0.035 (0.006)	7.26E-09
UA	rs2941484	T	HNF4G	8:75566533	0.049 (0.006)	3.91E-17
UA	rs3741414	T	INHBC	12:57450266	–0.071 (0.007)	9.79E-22
UA	rs642803	T	OVOL1	11:65793149	–0.043 (0.005)	4.51E-14
UA	rs653178	T	ATXN2	12:11156995	–0.036 (0.005)	2.45E-10
UA	rs6598541	A	IGF1R	15:98727906	0.044 (0.006)	5.20E-13
UA	rs675209	T	NA	6:7101851	0.063 (0.006)	1.38E-21
UA	rs6770152	T	NA	3:53066198	–0.048 (0.006)	2.66E-16
UA	rs7193778	T	NA	16:69529987	–0.047 (0.008)	2.36E-08
UA	rs7224610	A	HLF	17:55287427	–0.038 (0.006)	4.74E-11
UA	rs729761	T	POLR1C	6:43836834	–0.046 (0.006)	3.05E-12
UA	rs7654258	T	LOC10798626	4:10402888	–0.099 (0.010)	1.09E-19
Gout	rs1481012	A	ABCG2	4:88117930	–0.550 (0.046)	2.05E-32
Gout	rs4475146	A	SLC2A9	4:9945032	–0.470 (0.044)	4.11E-26

The SNP information was obtained from NCBI (https://www.ncbi.nlm.nih.gov/snp/). The position of the SNP was corresponded to GRCh38.p13. NA, not available; SNP, single nucleotide polymorphism; EA, effect allele; SE, standard error.

Table 2

Heterogeneity and pleiotropy tests of instrument effects

Serum UA							Gout
Method	Q	Q_df	p	Egger_intercept	SE	p	Q	Q_df	p	Egger_intercept	SE	p
AD as the outcome
IVW	23.057	24	0.516	–	–	–	0.587	1	0.444	–	–	–
MR Egger	19.926	23	0.646	–0.011	0.006	0.090	–	–	–	–	–	–
AD as the exposure
IVW	4.144	3	0.246	–	–	–	0.274	3	0.965		–	–
MR Egger	3.717	2	0.156	–0.015	0.032	0.679	0.190	2	0.909	0.041	0.140	0.799

UA, uric acid; AD, Alzheimer’s disease; IVW, Inverse variance weighted; MR, Mendelian randomization; SE, standard error.

Genetically predicted serum UA levels and risk of AD

A statistically significant causal effect of genetically predicted serum UA levels on AD was revealed using 25 SNPs (OR_IVW = 1.09, 95% CI = 1.01–1.19, p = 0.031; Fig. 2, Supplementary Figure 1). However, the p value of 0.031 failed to pass the multiple corrected p value of 0.025. Sensitivity analyses, including MR-Egger (OR_MR - Egger = 1.22, 95% CI = 1.06–1.42, p = 0.014), weighted median (OR_{weightedmedian} = 1.18, 95% CI = 1.05–1.33, p = 0.006) and weighted mode (OR_weightedmode = 1.20, 95% CI = 1.07–1.35, p = 0.005) methods confirmed the causal effect. No evidence of heterogeneity (p for MR Egger = 0.646, p for IVW = 0.516; Table 1) or directional pleiotropy (intercept: –0.011, p = 0.090; p for MR-PRESSO global test = 0.460) were found for this association. However, the leave-one-out analysis identified three outliers, rs2078267, rs2231142, and rs11722228 significantly driving the overall direction (Supplementary Figure 2). The F statistics of 151.23 indicated that the result was less likely to be affected by weak instrumental bias.

Fig. 2

Mendelian randomization results of serum UA level, gout and AD. A summary of 25 SNPs (rs10761587, rs11264341, rs1165151, rs1171614, rs11722228, rs1178977, rs1260326, rs1394125, rs1471633, rs17050272, rs1825043, rs2078267, rs2231142, rs2307394, rs2941484, rs3741414, rs642803, rs653178, rs6598541, rs675209, rs6770152, rs7193778, rs7224610, rs729761, and rs7654258) was included in the UA-AD association, and 2 SNPs (rs1481012, and rs4475146) were used to examine the association between gout and AD (detailed information of the SNPs can be found in Table 1). Genetic proxies for serum UA concentration (OR_IVW = 1.09) were related with an increased risk of AD. This causal effect was further confirmed using sensitivity analyses including MR-Egger (OR: 1.22), weighted median (OR: 1.18) and weighted mode (OR: 1.20) methods. Supportive causal effect of genetically determined gout on AD risk was also demonstrated (OR_IVW = 1.05). Bold fonts represent significant results. UA, uric acid; SNP, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval; MR, Mendelian randomization; AD, Alzheimer’s disease; IVW, inverse variance weighted.

Genetically predicted gout and risk of AD

Supportive causal association between genetically determined gout and AD risk was found using two SNPs (OR_IVW = 1.05, 95% CI = 1.00–1.11, p = 0.057; Fig. 2, Supplementary Figure 3), whereas no sensitivity analysis could be performed due to the limited SNPs (N = 2). There was no evidence of heterogeneity in the causal effect estimates (p for IVW = 0.444; Table 1). The F statistic was 128.53.

Genetically determined AD and serum UA levels and gout

No causal effects of AD on serum UA levels (4 SNPs) were found (OR_IVW = 0.97, 95% CI = 0.91–1.04, p = 0.412; Fig. 3, Supplementary Figure 4). The Cochran Q statistic (p for MR Egger = 0.156, p for IVW = 0.246) indicated no significant heterogeneity across instrument SNP effects (Table 1). The MR-PRESSO global test indicated no pleiotropy (p = 0.312). No single SNP driven the above associations (Supplementary Figure 5). The F statistics for the association was 49.19.

Fig. 3

Mendelian randomization results of AD and serum UA level, gout. Four SNPs, rs11218343, rs11767557, rs1582763, and rs3851179 were included. No causal effects of AD on serum UA levels (OR_IVW = 0.97) and gout risk (OR_IVW = 0.25) were found. The summary information list of included SNPs can be found in Supplementary Tables 1and 2 . UA, uric acid; SNP, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval; MR, Mendelian randomization; AD, Alzheimer’s disease; IVW, inverse variance weighted.

AD was also not casually related with gout according to the present result (OR_IVW = 0.25, 95% CI = 0.67–1.29, p = 0.653; Fig. 3, Supplementary Figure 6). No heterogeneity was detected (p for MR Egger = 0.909, p for IVW = 0.965). No evidence of pleiotropy was found (intercept: 0.041, p = 0.799; p for MR-PRESSO global test = 0.968). No single SNP driven the above associations (Supplementary Figure 7). The F statistics for the association was 49.19.

DISCUSSION

Using large summary-level GWAS data, this is the first bidirectional two-sample MR study to examine the causal associations between serum UA levels and gout with AD risk. MR analyses supported a causal risky role of lifelong genetic increase in serum UA for the risk of AD (Fig. 4), instead, no causal effect of AD on serum UA levels was demonstrated. A suggestive causal risk effect of gout on the increased risk of AD was also found, however, no inverse causality was identified. The consistency of results across MR methods strengthened our inference of causality, thus our results were less likely to be driven by genetic pleiotropy and heterogeneity.

Fig. 4

Summary findings for the MR analysis of serum UA levels, gout, and risk of AD. A causal relationship between elevated serum UA level and AD risk was revealed. However, no reverse causal effects of AD on serum UA levels and gout risk were found. UA, uric acid; OR, odds ratio; AD, Alzheimer’s disease.

In recent decades, the association between dementia and serum UA levels has attracted extensive attention, but there was still a lack of clear evidence. A meta-analysis of 21 case-control studies demonstrated that serum UA concentrations were evidently lower in AD patients than in cognitively normal controls [41]. Two large-scale prospective studies revealed a dementia risk reduction in patients with higher serum UA levels [12, 42]. In a cohort of 1,462 female participants followed up over 30 years, researchers found that UA levels in serum were negatively associated with the risk of incident dementia, including AD [43]. All of the above risk reduction of dementia associated with increasing serum UA levels highlighted some possible beneficial function of UA on dementia. Conversely, a significant association between hyperuricemia with worse cognitive function and brain white matter atrophy was demonstrated among 814 patients from the Rotterdam Scan Study [44]. In the prospective French 3C Study followed for 10.1 years, participants with the highest level of serum UA were at higher risk of developing incident dementia [13]. However, there was no significant association with AD. The reasons for the inconsistent results of those studies are equivocal. It might be due to the differences in the analytical populations (i.e., ethnicity, age range, and gender), dementia diagnosis, study design (i.e., prospective, retrospective, and case control studies) and the confounding factors adjusted in the statistical models in previous studies [13].

Most of the existing investigations concerning this topic mainly focused on the effect of UA on the risk of AD occurrence and cognitive decline; however, few explorations on the other aspects were performed, such as the age of AD onset, and disease severity or progression. Studies have suggested that the high UA level was associated with slower cognitive decline in AD patients [45, 46], and could attenuate the effects of amyloid-β (Aβ) and tau on decline of cognition in female individuals [27]. Conversely, some other studies have shown that UA had an adverse influence on cognition [47 –49]. Extracellular Aβ aggregation and intracellular neurofibrillary tangles (NFT) constitute two hallmarks of AD, accumulated silently decades before the clinical symptoms become evident [50, 51]. Serum UA has reported to be related with AD-related core pathological changes. It could aggravate the detrimental effects of Aβ deposition on neuronal loss supported by an experimental study in vitro. Our research group has recently found that the elevation of serum UA was related with cerebrospinal fluid (CSF) Aβ levels in preclinical stage of AD within a cohort of northern Han Chinese [52]. However, the role of UA in tauopathy has not yet been fully evaluated and the conclusions were inconsistent. Tommaso et al. observed that patients with different tauopathies had significantly lower serum UA levels compared to healthy controls [53]. Inconsistent with this finding, a previous study of patients with primary tauopathies reported normal content of UA [54]. Given the close association between UA and AD core pathology, we considered whether UA could be a viable biomarker for AD. Regrettably, barely any studies explored whether serum UA could be a potential AD biomarker, partly due to the extreme controversial results between the two traits, the dual effects and gender-specific effects of UA, and the uncontrolled confounding, e.g., the nutrition and comorbidity status. Only after this relationship has been confirmed by more studies can clinical trials be designed from this point to using serum UA as a reliable AD biomarker, and further a therapeutic target [55].

One possible explanation for the relationship between elevated serum UA levels and AD is that it might be mediated by the underlying cerebrovascular diseases [56 –58]. UA can cause the pathogenesis of cerebrovascular diseases by increasing endothelial cell activation and platelet adhesion [59], eventually elevated the risk of dementia. Furthermore, the most well-known hypothesis that the beneficial effects of serum UA on neurodegeneration could be explained by the antagonism between the antioxidant function of UA and the oxidative stress status was partly questioned by Hershfield et al.’ study, where no changes in oxidative stress markers were observed in patients who received preparations that reduces serum UA concentration [60]. Contrary to this widely accepted view of the antioxidant function of UA, in vitro studies even found that UA could increase oxidative stress and enhance the neurotoxic effects of Aβ in neurons [8]. On the consequence, UA and Aβ might synergistically promote AD initiation and subsequent neurodegenerative processes. In terms of underlying mechanisms, the genes identified by large GWASs were not only associated serum UA levels, but also involved in altering the risk of AD occurrence. Pathways that they may be involved in include innate immune system (TRIM46 [61], ATXN2 [62]), regulation of gene transcription and expression (BAZ1B [63], HNF4G [64], HLF [65]), metabolic processes (UBE2Q2 [66], SLC2A9 [67], GCKR [68]), and regulation of microglia function (IGF1R [69]). In general, the relationship between serum UA concentration and dementia is still unclear, and the underlying mechanisms are very complex. Our present results only provided one possible direction, thus more research is needed to support the finding and validate it.

Systematic biases, such as reverse causation and unmeasured confounders exist in observational design, but MR design can circumvent these biases. Over the last few years, four MR studies attempted to uncover the causality between UA with AD and cognitive performances [24 –27]. All of the three MR studies for AD [24 , 27] obtained ORs > 1, implying that higher exposure of UA confers slightly elevated risk of AD, rather than neuroprotection. Another MR analysis used 28 genetic variants, providing suggestive evidence of the association between higher genetically predicted serum UA and cognitive performances [25]. These MR analyses regarding on AD risk extracted IVs from the IGAP GWAS published in 2013. However, our analysis utilized the larger GWAS published recently [31] and more strict selection criteria, thus could provide more convincing results. To exclude pleiotropy, we conducted several sensitivity analyses (MR-Egger, weighted median, and weighted mode), which could provide unbiased causal effect estimates even when many of the genetic instruments were invalid. Causal relationships were supported by all of the four analyses. Moreover, we are the first to investigate the effect of AD on UA and gout risk, though no significant effects were found. Notably, this result should be interpreted cautiously, which may be due to the limited number of SNPs (N = 2) included in the analysis. Future large-scale GWASs would render more statistical power.

Limitations

Several limitations merit careful consideration. First, two-sample MR models have several general defects, including the assumption of the linear exposure-outcome correlation and bias due to genetic phenomena [70]. Second, MR measures the cumulative lifelong effect of UA-related genetic variants in serum, therefore the results should not be extrapolated to assume the effect of UA on AD at a specific time point or period, and the same goes for gout. Third, we cannot be certain for that the selected SNPs do not violate the exclusion-restriction assumption, particularly regarding the unmeasured confounding. However, the MR-PRESSO and Egger regression approach indicated that the results were robust to the violations of MR hypothesis. Furthermore, the consistency of estimates across multiple sensitivity analyses also increases the confidence in the presence of a true potential causal effect. Fourth, the use of two SNPs to instrument gout limited the statistical power, even with the very large AD case-control samples available. Larger genetic GWASs could provide more independent SNPs as IVs, which would further increase the power of MR studies and allow for more nuanced exploration of bias caused by genetic pleiotropy. Last, since the majority of included individuals are of European origin, the conclusions derived from this study are not necessarily applicable to other ethnicities. Investigations within other ethnicities are necessary.

In summary, our study assessed the association between lifelong genetically increased serum UA levels, gout, and the risk of developing AD. The findings demonstrated here suggested that increasing serum UA was positively and causally related with the risk of AD incidence. If our findings are validated in other study populations and other study approaches, it may have important implications for the treatment of AD dementia, as well as hyperuricemia and gout. Clearly and more importantly, more investigations exploring the underlying mechanisms of this association are required.

Footnotes

ACKNOWLEDGMENTS

This work was made possible by the generous sharing of GWAS summary statistics. We gratefully acknowledge the authors and participants of all GWAS from which we used summary statistics data. We thank Kunkle et al. for the AD GWAS summary results data, and Köttgen et al. for the UA and gout GWAS data. IGAP was made possible by the generous participation of the control subjects, the patients, and their families. The i–Select chips were funded by the French National Foundation on Alzheimer’s disease and related disorders. EADI was supported by the LABEX DISTALZ grant, Inserm, Institut Pasteur de Lille, Université de Lille 2, and the Lille University Hospital. GERAD was supported by the Medical Research Council (Grant no 503480),Alzheimer’s Research UK (Grant no 503176), the Wellcome Trust (Grant no 082604/2/07/Z), and German Federal Ministry of Education and Research (BMBF):Competence Network Dementia (CND) grant no 01GI0102, 01GI0711, 01GI0420.CHARGE was partly supported by the NIH/NIA grant R01 AG033193 and the NIAAG081220 and AGES contract N01–AG–12100, the NHLBI grant R01 HL105756, the Icelandic Heart Association, and the Erasmus Medical Center and Erasmus University. ADGC was supported by the NIH/NIA grants: U01 AG032984, U24 AG021886, U01AG016976, and the Alzheimer’s Association grant ADGC–10–196728. The investigators within above consortiums contributed to the design and implementation of the GWAS datasets and/or provided data but did not participate in analysis or writing of this report.

This study was supported by grants from the National Natural Science Foundation of China (82071201, 91849126), Shanghai Municipal Science and Technology Major Project (No.2018SHZDZX01), Research Start-up Fund of Huashan Hospital (2022QD002), Excellence 2025 Talent Cultivation Program at Fudan University (3030277001), and ZHANGJIANG LAB, Tianqiao and Chrissy Chen Institute, and the State Key Laboratory of Neurobiology and Frontiers Center for Brain Science of Ministry of Education, Fudan University.

Authors’ disclosures available online ().

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

References

Heemels

(2016) Neurodegenerative diseases. Nature 539, 179.

(2022) 2022 Alzheimer’s disease facts and figures. Alzheimers Dement 18, 700–789.

Jack

Jr. , Lowe

, Weigand

, Wiste

, Senjem

, Knopman

, Shiung

, Gunter

, Boeve

, Kemp

, Weiner

, Petersen

(2009) Serial PIB and MRI in normal, mild cognitive impairment and Alzheimer’s disease: Implications for sequence of pathological events in Alzheimer’s disease. Brain 132, 1355–1365.

Villemagne

, Burnham

, Bourgeat

, Brown

, Ellis

, Salvado

, Szoeke

, Macaulay

, Martins

, Maruff

, Ames

, Rowe

, Masters

(2013) Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: A prospective cohort study. Lancet Neurol 12, 357–367.

Butterfield

, Halliwell

(2019) Oxidative stress, dysfunctional glucose metabolism and Alzheimer disease. Nat Rev Neurosci 20, 148–160.

Tönnies

, Trushina

(2017) Oxidative stress, synaptic dysfunction, and Alzheimer’s disease. J Alzheimers Dis 57, 1105–1121.

Bowman

, Shannon

, Frei

, Kaye

, Quinn

(2010) Uric acid as a CNS antioxidant. J Alzheimers Dis 19, 1331–1336.

Desideri

, Gentile

, Antonosante

, Benedetti

, Grassi

, Cristiano

, Manocchio

, Selli

, Ippoliti

, Ferri

, Borghi

, Giordano

, Cimini

(2017) Uric acid amplifies Aβ amyloid effects involved in the cognitive dysfunction/dementia: Evidences from an experimental model}. J Cell Physiol 232, 1069–1078.

, Thorens

(2010) Uric acid transport and disease. J Clin Invest 120, 1791–1799.

10.

Tana

, Ticinesi

, Prati

, Nouvenne

, Meschi

(2018) Uric acid and cognitive function in older individuals. Nutrients 10, 975.

11.

, Xu

, Tan

, Andrieu

, Suckling

, Evangelou

, Pan

, Zhang

, Jia

, Feng

, Kua

, Wang

, Tan

, Li

, Hou

, Wan

, Tan

, Mok

, Tan

, Dong

, Touchon

, Gauthier

, Aisen

, Vellas

(2020) Evidence-based prevention of Alzheimer’s disease: Systematic review and meta-analysis of 243 observational prospective studies and 153 randomised controlled trials. J Neurol Neurosurg Psychiatry 91, 1201–1209.

12.

Euser

, Hofman

, Westendorp

, Breteler

(2009) Serum uric acid and cognitive function and dementia. Brain 132, 377–382.

13.

Latourte

, Soumaré

, Bardin

, Perez-Ruiz

, Debette

, Richette

(2018) Uric acid and incident dementia over 12 years of follow-up: A population-based cohort study. Ann Rheum Dis 77, 328–335.

14.

Richette

, Perez-Ruiz

, Doherty

, Jansen

, Nuki

, Pascual

, Punzi

, So

, Bardin

(2014) Improving cardiovascular and renal outcomes in gout: What should we target? Nat Rev Rheumatol 10, 654–661.

15.

, Dubreuil

, Zhang

, Neogi

, Rai

, Ascherio

, Hernán

, Choi

(2016) Gout and the risk of Alzheimer’s disease: A population-based, BMI-matched cohort study. Ann Rheum Dis 75, 547–551.

16.

Hong

J-Y

, Lan

T-Y

, Tang

G-J

, Tang

C-H

, Chen

T-J

, Lin

H-Y

(2015) Gout and the risk of dementia: A nationwide population-based cohort study. Arthritis Res Ther 17.

17.

Singh

, Cleveland

(2018) Gout and dementia in the elderly: A cohort study of Medicare claims. BMC Geriatr 18, 281.

18.

Walker

, Davey Smith

, Davies

, Martin

(2017) Mendelian randomization: A novel approach for the prediction of adverse drug events and drug repurposing opportunities. Int J Epidemiol 46, 2078–2089.

19.

Baumeister

, Karch

, Bahls

, Teumer

, Leitzmann

, Baurecht

(2020) Physical activity and risk of Alzheimer disease: A 2-sample mendelian randomization study. Neurology 95, e1897–e1905.

20.

Huang

, Zuber

, Matthews

, Elliott

, Tzoulaki

, Dehghan

(2020) Sleep, major depressive disorder, and Alzheimer disease: A Mendelian randomization study. Neurology 95, e1963–e1970.

21.

Andrews

, Goate

, Anstey

(2020) Association between alcohol consumption and Alzheimer’s disease: A Mendelian randomization study. Alzheimers Dement 16, 345–353.

22.

Larsson

, Traylor

, Malik

, Dichgans

, Burgess

, Markus

(2017) Modifiable pathways in Alzheimer’s disease: Mendelian randomisation analysis. BMJ 359, j5375.

23.

Andrews

, Fulton-Howard

, O’Reilly

, Marcora

, Goate

(2021) Causal associations between modifiable risk factors and the Alzheimer’s phenome. Ann Neurol 89, 54–65.

24.

Williams

, Hägg

, Pedersen

(2019) Circulating antioxidants and Alzheimer disease prevention: A Mendelian randomization study. Am J Clin Nutr 109, 90–98.

25.

Efstathiadou

, Gill

, McGrane

, Quinn

, Dawson

(2019) Genetically determined uric acid and the risk of cardiovascular and neurovascular diseases: A Mendelian randomization study of outcomes investigated in randomized trials. J Am Heart Assoc 8, e012738.

26.

Yuan

, Yang

, Rizzuto

(2018) Genetically determined serum uric acid and Alzheimer’s disease risk. J Alzheimers Dis 65, 1259–1265.

27.

Wang

, Meng

, Shen

, Ji

(2020) Impact of modifiable risk factors on Alzheimer’s disease: A two-sample Mendelian randomization study. Neurobiol Aging 91, 167.e111–167.e119.

28.

Davey Smith

, Hemani

(2014) Mendelian randomization: Genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 23, R89–98.

29.

Davies

, Holmes

, Davey Smith

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

30.

Köttgen

, Albrecht

, Teumer

, Vitart

, Krumsiek

, Hundertmark

, Pistis

, Ruggiero

, O’Seaghdha

, Haller

, et

al.

(2012) Genome-wide association analyses identify 18 new loci associated with serum urate concentrations. Nat Genet 45, 145–154.

31.

Kunkle

, Grenier-Boley

, Sims

, Bis

, Damotte

, Naj

, Boland

, Vronskaya

, van der Lee

, Amlie-Wolf

,et al. (2019) Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Abeta, tau, immunity and lipid processing. Nat Genet 51, 414–430.

32.

Bowden

, Davey Smith

, Burgess

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

33.

Burgess

, Butterworth

, Thompson

(2013) Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 37, 658–665.

34.

Hemani

, Bowden

, Davey Smith

(2018) Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet 27, R195–R208.

35.

Hartwig

, Davey Smith

, Bowden

(2017) Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol 46, 1985–1998.

36.

Bowden

, Del Greco

, Minelli

, Davey Smith

, Sheehan

, Thompson

(2017) A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization. Stat Med 36, 1783–1802.

37.

Bowden

, Del Grecos

, Minelli

, Zhao

, Lawlor

, Sheehan

, Thompson

, Davey Smith

(2019) Improving the accuracy of two-sample summary-data Mendelian randomization: Moving beyond the NOME assumption. Int J Epidemiol 48, 728–742.

38.

Verbanck

, Chen

, Neale

, Do

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

39.

Pichler

, Del Greco

M F

, Gögele

, Lill

, Bertram

, Do

, Eriksson

, Foroud

, Myers

; PD GWAS Consortium, Nalls

, Keller

; International Parkinson’s Disease Genomics Consortium; Wellcome Trust Case Control Consortium 2, Benyamin

, Whitfield

; Genetics of Iron Status Consortium, Pramstaller

, Hicks

, Thompson

, Minelli

(2013) Serum iron levels and the risk of Parkinson disease: A Mendelian randomization study. PLoS Med 10, e1001462.

40.

Burgess

, Thompson

, CRP CHD Genetics Collaboration (2011) Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40, 755–764.

41.

, Xu

, Hou

, Song

, Liu

, Chen

, Wang

, Li

(2016) Inverse association between serum uric acid levels and Alzheimer’s disease risk. Mol Neurobiol 53, 2594–2599.

42.

Cao

, Xu

, Yang

, Li

, Xu

, Zhang

, Wang

(2020) Associations of BMI and serum urate with developingdementia: A prospective cohort study. J Clin Endocrinol Metab 105.

43.

Scheepers

, Jacobsson

LTH

, Kern

, Johansson

, Dehlin

, Skoog

(2019) Urate and risk of Alzheimer’s disease and vascular dementia: A population-based study. Alzheimers Dement 15, 754–763.

44.

Verhaaren

, Vernooij

, Dehghan

, Vrooman

, de Boer

, Hofman

, Witteman

, Niessen

, Breteler

, van der Lugt

, Ikram

(2013) The relation of uric acid to brain atrophy and cognition: The Rotterdam Scan Study. Neuroepidemiology 41, 29–34.

45.

Beydoun

, Canas

, Dore

, Beydoun

, Rostant

, Fanelli-Kuczmarski

, Evans

, Zonderman

(2016) Serum uric acid and its association with longitudinal cognitive change among urban adults. J Alzheimers Dis 52, 1415–1430.

46.

Lin

, Zheng

, Joseph Schoepf

, Varga-Szemes

, Savage

, Wang

, Zhang

, Lu

, Zhang

(2019) Uric acid has different effects on spontaneous brain activities of males and females: A cross-sectional resting-state functional MR imaging study. Front Neurosci 13, 763.

47.

Cicero

, Desideri

, Grossi

, Urso

, Rosticci

, D’Addato

, Borghi

(2015) Serum uric acid and impaired cognitive function in a cohort of healthy young elderly: Data from the Brisighella Study. Intern Emerg Med 10, 25–31.

48.

Zhang

, Tang

, Hu

, Wang

, Xu

(2020) Uric acid is associated with cognitive impairment in the elderly patients receiving maintenance hemodialysis-A two-center study. Brain Behav 10, e01542.

49.

Suzuki

, Koide

, Fujii

, Yamazaki

, Tsuji

, Iwata

(2016) Elevated serum uric acid levels are related to cognitive deterioration in an elderly Japanese population. Dement Geriatr Cogn Dis Extra 6, 580–588.

50.

Ballard

, Gauthier

, Corbett

, Brayne

, Aarsland

, Jones

(2011) Alzheimer’s disease. Lancet 377, 1019–1031.

51.

Aguzzi

, O’Connor

(2010) Protein aggregation diseases: Pathogenicity and therapeutic perspectives. Nat Rev Drug Discov 9, 237–248.

52.

L-L

, Ma

Y-H

, Bi

Y-L

, Sun

F-R

, Hu

, Hou

X-H

, Xu

, Shen

X-N

, Dong

, Tan

, Yang

J-L

, Yu

J-T

, Zhu

L-Q

(2021) Serum uric acid may aggravate Alzheimer’s disease risk by affecting amyloidosis in cognitively intact older adults: The CABLE Study. J Alzheimers Dis 81, 389–401.

53.

Schirinzi

, Di Lazzaro

, Colona

, Imbriani

, Alwardat

, Sancesario

, Martorana

, Pisani

(2017) Assessment of serum uric acid as risk factor for tauopathies. J Neural Transm (Vienna) 124, 1105–1108.

54.

Brody

, Litvan

, Warner

, Riley

, Hall

, Kluger

, Shprecher

, Cunningham

(2016) Relationship between uric acid levels and progressive supranuclear palsy. Mov Disord 31, 663–667.

55.

Qiao

, Chen

, Liang

, Luo

, Wu

, Xu

(2021) The influence of serum uric acid level on Alzheimer’s disease: A narrative review. BioMed Res Int 2021, 5525710.

56.

Feig

, Kang

, Johnson

(2008) Uric acid and cardiovascular risk. N Engl J Med 359, 1811–1821.

57.

Vannorsdall

, Jinnah

, Gordon

, Kraut

, Schretlen

(2008) Cerebral ischemia mediates the effect of serum uric acid on cognitive function. Stroke 39, 3418–3420.

58.

Borghi

, Rodriguez-Artalejo

, De Backer

, Dallongeville

, Medina

, Nuevo

, Guallar

, Perk

, Banegas

, Tubach

, Roy

, Halcox

(2018) Serum uric acid levels are associated with cardiovascular risk score: A} analysis of the EURIKA study. Int J Cardiol 253, 167–173.

59.

Kanellis

, Kang

(2005) Uric acid as a mediator of endothelial dysfunction, inflammation, and vascular disease. Semin Nephrol 25, 39–42.

60.

Hershfield

, Roberts

, 2nd, Ganson

, Kelly

, Santisteban

, Scarlett

, Jaggers

, Sundy

(2010) Treating gout with pegloticase, a PEGylated urate oxidase, provides insight into the importance of uric acid as an antioxidant}. Proc Natl Acad Sci U S A 107, 14351–14356.

61.

Gibson

, Russ

, Clarke

, Howard

, Hillary

, Evans

, Walker

, Bermingham

, Morris

, Campbell

, Hayward

, Murray

, Porteous

, Horvath

, Lu

, McIntosh

, Whalley

, Marioni

(2019) A meta-analysis of genome-wide association studies of epigenetic age acceleration. PLoS Genet 15, e1008104.

62.

Orrù

, Steri

, Sole

, Sidore

, Virdis

, Dei

, Lai

, Zoledziewska

, Busonero

, Mulas

, Floris

, Mentzen

, Urru

, Olla

, Marongiu

, Piras

, Lobina

, Maschio

, Pitzalis

, Urru

, Marcelli

, Cusano

, Deidda

, Serra

, Oppo

, Pilu

, Reinier

, Berutti

, Pireddu

, Zara

, Porcu

, Kwong

, Brennan

, Tarrier

, Lyons

, Kang

, Uzzau

, Atzeni

, Valentini

, Firinu

, Leoni

, Rotta

, Naitza

, Angius

, Congia

, Whalen

, Jones

, Schlessinger

, Abecasis

, Fiorillo

, Sanna

, Cucca

(2013) Genetic variants regulating immune cell levels in health and disease. Cell 155, 242–256.

63.

Geiger

, Rieckmann

, Wolf

, Basso

, Feng

, Fuhrer

, Kogadeeva

, Picotti

, Meissner

, Mann

, Zamboni

, Sallusto

, Lanzavecchia

(2016) L-arginine modulates T cell metabolism and enhances survival and anti-tumor activity. Cell 167, 829–842.e813.

64.

Chen

, Toke

, Luo

, Vasoya

, Fullem

, Parthasarathy

, Perekatt

, Verzi

(2019) A reinforcing HNF4-SMAD4 feed-forward module stabilizes enterocyte identity. Nat Genet 51, 777–785.

65.

Lehnertz

, Chagraoui

, MacRae

, Tomellini

, Corneau

, Mayotte

, Boivin

, Durand

, Gracias

, Sauvageau

(2021) HLF expression defines the human hematopoietic stem cell state. Blood 138, 2642–2654.

66.

Yang

, Zhang

, Cao

, Xu

, Chen

, Wang

, Li

, Zhang

, Xie

, Fang

, Xu

(2021) Circular RNA UBE2Q2 promotes malignant progression of gastric cancer by regulating signal transducer and activator of transcription 3-mediated autophagy and glycolysis. Cell Death Dis 12, 910.

67.

Mandal

, Mount

(2019) Interaction between ITM2B and GLUT9 links urate transport to neurodegenerative disorders. Front Physiol 10, 1323.

68.

Zahedi

, Akbarzadeh

, Sedaghati-Khayat

, Seyedhamzehzadeh

, Daneshpour

(2021) GCKR common functional polymorphisms are associated with metabolic syndrome and its components: A 10-year retrospective cohort study in Iranian adults. Diabetol Metab Syndr 13, 20.

69.

Jian

, Wei

, Mo

, Li

, Liang

, Chen

, Zou

, Meng

, Liu

, Zou

(2021) Microglia mediate the occurrence and development of Alzheimer’s disease through ligand-receptor axis communication. Front Aging Neurosci 13, 731180.

70.

Gao

, Patel

, Michailidou

, Peters

, Gong

, Schildkraut

, Schumacher

, Zheng

, Boffetta

, Stucker

, Willett

, Gruber

, Easton

, Hunter

, Sellers

, Haiman

, Henderson

, Hung

, Amos

, Pierce

, Lindström

, Kraft

(2016) Mendelian randomization study of adiposity-related traits and risk of breast, ovarian, prostate, lung and colorectal cancer. Int J Epidemiol 45, 896–908.