Serum urate or uric acid is an end product of purine metabolism (1). High serum urate or hyperuricemia is the most critical risk factor for gout (2). Due to the spread of Western lifestyles and diets associated with excessive meat and alcohol consumption, hyperuricemia has become gradually more common (~20.0%; US; 2016) (3). Moreover, the extensive associations of serum urate with age-related diseases, such as cardiovascular diseases (4, 5), hypertension (6), obesity (6), type 2 diabetes mellitus (7), and metabolic syndrome (8), have been implied. The effects of urate on cellular functions and signaling events, such as promoting the production of reactive oxygen species (ROS) and activating inflammatory signaling pathways, have indicated the possible involvement in systematic inflammation (9–12). Due to its complicated connections within multiple diseases, the role of urate on multiple organ systems that may be facilitated by a perturbation in circulation has been a matter of discussion. Previous studies found the associations between serum urate and serum inflammation markers, such as C-reactive protein (CRP), IL-6, and TNF-α (13). Co-occurrence of high serum urate and high CRP is associated with elevated mortality (14, 15). However, whether these associations are causal remains unclear due to potential biases such as residual confounding and reverse causality.

Telomeres are nucleoprotein structures at the ends of chromosomes that protect linear chromosomes against damage by endogenous nucleases and erroneous recognition as unrepaired chromosomal breaks (16). Due to their critical functions in the maintenance of chromosomal structure and stability (17) and the determination of cell proliferation (18), telomere length has been considered an indicator of the biological age of organisms and a biomarker of aging, stress, and survival (19–21). Telomere shortening is associated with a series of age-related deteriorations and diseases, such as cardiovascular (22) and metabolic diseases (23). Interestingly, a significant negative association between serum urate and peripheral blood aging markers, including leukocyte telomere length (LTL) and mitochondrial DNA copy number (mtDNAcn) (24), has indicated that serum urate may play a role in promoting organismal aging by telomere shortening. Although reduced telomere length and elevated serum urate levels both come with aging, a recent Mendelian randomization (MR) study does not support the casual effect of serum urate on telomere shorten (25). However, the casual association between them remains elusive because of the limited numbers of single nucleotide polymorphisms (SNPs) employed as instrumental variables (IVs) in the previous study.

Here, we used an MR design to overcome the limitations of traditional observational studies and determine whether there are causal associations between serum urate and LTL and serum soluble makers of inflammation (CRP, IL-6, TNF-α, and IGF-1) (26). To detect whether serum urate causally affects LTL, more importantly, we employed a ~4-fold larger number of SNPs as IVs than the previous MR study (25). Confounding and reverse causality can be avoided in MR analysis. This study provides reliable insight into the causal associations between serum urate and the markers of aging.

To assess the causal association of genetically predicted urate level on aging-related biomarkers, we implemented an MR approach, assuming that (i) the SNPs used as instrumental variables (IVs) are robustly associated with urate level; (ii) the SNPs are not associated with any confounder of exposure-outcome associations; (iii) the SNPs exert effects only through urate level rather than via other pathways. For SNPs to instrument urate level, we obtained the results from a published meta-analysis of GWASs (27). Each GWAS in this study has been approved by relevant ethical review committees, all the participants gave written informed consent. Summary-level dataset for outcomes could be obtained from the UK Medical Research Council Integrative Epidemiology Unit (MRCIEU) Open GWAS Project database (http://gwas.mrcieu.au.uk). Therefore, no additional ethical approval is required for our study.

Detailed information of selected 101 SNPs to instrument urate level was summarized in Supplementary Table 1 . The F-statistics of these selected SNPs ranged from 62.9 to 5561.5, which were much higher than the conventional threshold of 10, suggesting a low possibility of weak-instrument bias. Based on a sample size of 472,174 individuals and the SNPs explaining 10.21% of the phenotypic variance, our study had 100% power to detect an OR of 1.20 (β=-0.2231, Table 1 ). The results of our MR study suggested an inverse association between urate level and LTL. The scatter plot of the association between genetically predicted urate level and LTL was shown in Figure 1A , where the slope of each fitted line represented the combined effect using each MR analysis method. Using the IVW method, genetically predicted urate level was negatively associated with LTL (0.040 SD higher urate per 1-SD shorter LTL, P=4.37×10^-8; Table 1 ). The inverse causal association was also supported by MR results using MR-Egger method and weighted median method ( Figure 1A and Table 2 ). As shown in Figure 1 , the MR-Egger intercept was close to zero (MR-Egger intercept=0.00018), and the MR-Egger regression analysis revealed no horizontal pleiotropy (P_-pleio=0.76). We observed significant heterogeneity across different SNPs (P_-het=1.34×10^-15), thus the random-effect model was used (forest plot shown in Figure 1B ). The MR-PRESSO approach identified four outlying SNPs, but the P value for distortion test (P=1.00) suggested that there was no significant difference between the original estimate and the outlier-corrected estimate of the causal effect. The leave-one-out sensitivity analysis also supported the robustness of the conclusion ( Supplementary Figure 1 ).

In terms of inflammation markers, including serum levels of CRP, IL-6, TNF-α, and IGF-1, none of our analyses suggested a causal effect of urate level ( Table 2 ). Based upon a sample size of 3,301 individuals and the phenotypic variance explained to 11.24%, our study had 99%/94% power to detect an OR of 0.80/1.20 ( Table 1 ).

Although basic research and clinical investigations have strongly indicated the implications of urate in organismal aging, its causal association with aging is not determined. In conflict with the previous MR study which reported that serum urate has no significant causal effect on LTL (25), we found that serum urate levels are reversely associated with LTL. The possible reason is that we employed a ~4-fold larger number of SNPs as IVs than the previous MR study. Moreover, the sample size of the GWAS we used in this study (27) is 4-fold larger than the previous GWAS studies (29, 49, 50). The post hoc power calculation suggested that our study had 100% power to detect an OR of 0.80/1.20, while the previous MR did not.

Since serum urate increases with age (51, 52), this result suggests that increased serum urate results in telomere shortening during aging. Telomere shortening is a well-documented hallmark of both cellular senescence and organismal aging, and is a critical mechanism of age-related pathologies and mortality (53). It happens in each division as a normal cellular process in the dividing cells (54). Moreover, telomere shortening occurs due to oxidative damage and other end processing events in both dividing and non-dividing cells (20). The effects of urate on cellular stress and ROS production may be the mechanism fascinating the urate-induced telomere shortening found in this study. ROS are produced simultaneously with the formation of intracellular urate by xanthine oxidases (XO) (55). Enhanced activity of XO promotes intracellular urate production. Moreover, elevated extracellular urate concentration aggravates the urate flux into cells. These events cause the accumulation of intracellular urate that aggravates ROS production (56, 57), mitochondrial damage (58), endoplasmic reticulum stress (59), and apoptosis (56). These cellular events cause telomere shortening alone or in an interactive manner; in turn, telomere shortening further aggravates cell stress, forming a closed loop (20, 60).

The role of intracellular urate has been demonstrated that it can trigger inflammation and ROS. However, we did not find a causal effect of serum urate on the markers of systematic inflammation. Previous studies have observed the associations between serum urate and serum inflammation indicators (13, 61, 62). Our study indicates that these associations are not causal. The possible interpretation of these conflict findings may be the paradox of oxidant versus anti-oxidant properties of urate in multiple systems (9) and the net effect of urate on system is variable depending on various conditions, such as genetics and disease progress.

MR studies rely on strong assumptions, and biased and misleading results could be yielded if those assumptions fail (63). In our current study, a total of 114 SNPs significantly associated with serum urate level were obtained from the largest GWAS meta-analysis, which was in line with the Relevance assumption. We then performed a comprehensive literature search to identify potential confounders that affect the exposure-outcome association. Only physical activity was deemed as a confounder through the literature search, but none of our selected SNPs was significantly associated with any physical activity-associated phenotype. It should be noted, if a particular confounder was simply not considered as a confounder in the analysis, its effect could not be accounted for. We made an effort to reduce the possibility that the Independence assumption was violated. Furthermore, we used three robust analysis methods (weighted median method, MR-Egger regression, and MR-PRESSO) as sensitivity analysis, since robust methods can provide valid causal inferences under weaker assumptions than the standard IVW method (44). Strikingly, causal estimates from all methods were similar ( Table 1 ), and the concordance of estimates from different methods made the causal claim more credible.

This study has several notable strengths. First, it is the first study revealing the causal association of serum urate with telomere length and highlighting the value of telomere length as a reliable serum marker reflecting urate-related aging. Second, an integrated and comprehensive MR analysis was conducted to support the robustness of causal estimates. A series of sensitivity analyses was also employed to ensure the consistency of causal association. Third, the number of SNPs employed as IVs is relatively large, enhancing the reliability of the conclusions.

Some limitations in this study need to be considered. First, the participants involved in the study were all of European descent, which may restrict the generalizability of the findings. Second, since a linear relationship between exposure and outcome was assumed in the MR design, the potential non-linear association of serum urate with leukocyte telomere length and inflammation markers could not be evaluated. Finally, other aging-related markers, such as immune parameters and indices of epigenetic age, were not included in this study.

Serum urate levels are negatively associated with telomere length but are not associated with inflammation markers. Telomere length may be a critical and reliable marker that reflects urate-related organismal aging and may be a mechanism in the age-related pathologies and mortality caused by hyperuricemia.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

ZL: conception and design, data analysis, and draft writing. JC: draft writing. JZ: draft writing, and revision. All authors approved the final manuscript.