Associations of serum uric acid, risk of atherosclerotic cardiovascular disease, and mortality: results from NHANES

Abstract

Background

Atherosclerotic cardiovascular disease (ASCVD) has long been recognized as a significant contributor to mortality rates, holding a prominent position in the hierarchy of causes of death. Nevertheless, the presence of a causal relationship between serum uric acid (SUA) and the risk of ASCVD, as well as mortality rates, remains unclear.

Methods

We initially conducted a comprehensive cohort study utilizing data sourced from National Health and Nutrition Examination Survey (NHANES) 1999–2018 to investigate the specific correlation between SUA levels and ASCVD. Then, we subsequently examined the link between SUA levels and all-cause and cardio-cerebrovascular mortality among ASCVD individuals.

Results

We identified a U-shaped relationship between SUA levels and the risk of ASCVD in all participants (inflection point at 5.399, p value = 0.014). Similarly, SUA levels showed U-shaped trends with all-cause mortality (inflection point at 5.748, p value < 0.0001) and cardio-cerebrovascular mortality (inflection point at 5.936, p value < 0.0001), respectively.

Conclusions

Our findings demonstrate a U-shaped association between SUA levels and the risks of ASCVD, all-cause mortality, and cardio-cerebrovascular mortality. However, further research is needed to better understand how SUA affects ASCVD and its underlying mechanisms.

Introduction

Atherosclerotic cardiovascular disease (ASCVD) has long been a prominent contributor to mortality rates in the United States and globally, occupying a leading position [1]. Coronary heart disease (CHD), stroke, and hypertension reign as the primary culprits behind both deaths and disabilities associated with ASCVD [2]. It was estimated that ASCVD accounts for a staggering 17.9 million fatalities, with myocardial infarction (MI) and stroke being the leading causes, encompassing 85% of these occurrences [1, 3]. ASCVD significantly diminishes the quality of life for affected individuals while subjecting them to heightened risks of death and disability [4]. Hence, the escalating mortality rates necessitate the urgent pursuit of innovative prevention and treatment strategies for ASCVD, alongside the identification of novel risk factors, which might ultimately contribute to alleviating the severity of this ailment [5]. Numerous studies have pinpointed a range of risk factors in the development of ASCVD, including gender, age, smoking, hypertension, hyperglycemia, and obesity [6]. However, recent studies have underscored the significance of uncovering new risk factors for ASCVD within the general population, as there exists a subset of patients who may develop the condition despite the absence of these traditional risk factors [7].

Serum uric acid (SUA), synthesized by xanthine oxidase, serves as the ultimate byproduct of purine catabolism. It is primarily eliminated from the body through urinary and fecal excretion [8]. SUA possesses notable antioxidant properties, playing a pivotal role as an endogenous antioxidant within the body. However, it has also been demonstrated to exert diverse pro-inflammatory, pro-oxidant, and vasoconstrictive effects [9]. Substantial elevation of SUA levels can lead to potential pathological ramifications. Prior research has established that elevated levels of SUA serve as a reliable marker for cardiovascular risk, with a robust correlation persisting even after adjusting for potential confounding variables [10,11,12]. Accumulating evidence suggests that elevated SUA levels contribute to the development of ASCVD and several cardiovascular risk factors [13,14,15]. Hyperuricemia (HUA) results from high SUA levels due to excessive uric acid production or reduced excretion, leading to its accumulation in the body. [16]. Epidemiological studies have consistently indicated a potential association between hyperuricemia and hypertension, metabolic syndrome, cardiovascular disease, and cerebrovascular disease. Numerous investigations have established higher SUA levels as a predictive factor for cardiovascular and all-cause mortality within the general population [17, 18]. Nevertheless, conflicting results have emerged from other studies, suggesting that higher SUA levels do not necessarily predict the onset of cardiovascular disease or all-cause mortality [19]. In addition, there exists evidence that lower SUA levels are also linked to accelerated mortality. Consequently, assessment of SUA levels as an independent risk factor for cardiovascular disease or all-cause mortality has yielded disparate outcomes [20]. The precise association between SUA levels and ASCVD remains unclear. Unraveling this association may provide novel insights into managing SUA levels in ASCVD individuals, particularly regarding intensive anti-uric acid therapy for ASCVD individuals afflicted with hyperuricemia.

To investigate the specific correlation between SUA levels and ASCVD, we initially conducted a comprehensive cohort study utilizing data sourced from NHANES 1999–2018. This extensive dataset encompasses a substantial sample of US adults, representative of the nation’s diverse socioeconomic and ethnic composition. Within this population, we subsequently examine the link between SUA levels and all-cause mortality and cardio-cerebrovascular mortality among individuals afflicted with ASCVD. The identification of such an association not only holds promise for advancing the prevention and treatment of ASCVD but also offers valuable insights into effectively managing SUA levels in ASCVD individuals, particularly in guiding the implementation of intensive anti-uric acid therapy for individuals with hyperuricemia.

Materials and methods

Study design

The National Health and Nutrition Examination Survey (NHANES) conducted by the National Center for Health Statistics (NCHS) was a comprehensive survey aimed at assessing the health and nutritional status of individuals in the United States. In this study, we performed both cross-sectional and prospective cohort analyses using data from NHANES participants between 1999 and 2018, who were 20 years of age or older, excluding those with a history of cancer. Figure 1 illustrates the study population, which excluded individuals with incomplete information regarding ASCVD, SUA, and covariates. The reporting of our findings adhered to the guidelines outlined in Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) for cross-sectional and prospective cohort studies. It was important to note that our study utilized publicly available data and was granted exempt status from the Ethics Committee of Shanghai Changhai Hospital.

Fig. 1

Flow chart diagram of study selection

Assessment of ASCVD

Within the medical conditions section of the household questionnaires conducted through home interviews, participants were asked the following question: “Has a doctor or other health professional ever informed that have had a coronary heart disease/angina, also known as angina pectoris/heart attack (also called myocardial infarction)/stroke?” Participants who responded positively to this question were categorized as having a history of ASCVD.

Definition of exposure

Participants who were 12 years old and below were included in the SUA examination as part of NHANES 1999–2018. The Laboratory Method Files section on the NHANES website (https://www.cdc.gov/nchs/nhanes) provided a detailed explanation of the laboratory methods employed. In this study, SUA levels were categorized into four groups based on their clinical relevance and previous research findings [21]. Group 1 comprised individuals with SUA levels below 6.0 mg/dL, Group 2 included those with levels between 6.0 and 6.7 mg/dL, Group 3 encompassed individuals with levels between 6.8 and 9.0 mg/dL, and Group 4 consisted of participants with SUA levels above 9.0 mg/dL.

Other variables of interest

To address potential biases in our study, we collected a diverse range of covariates to ensure comprehensive coverage of demographic factors, including age, gender, race, education level, marital status, and annual household income. We also accounted for various risk factors such as body mass index (BMI), smoking status, alcohol consumption, physical activity level, total energy intake, and adherence to the healthy eating index- 2015 (HEI- 2015). Furthermore, we considered the presence of comorbidities, including hypertension, diabetes mellitus, hyperlipidemia, menopausal status and chronic kidney disease (Supplemental Table 1).

It was important to note that certain medications, such as cyclosporine, niacin, diuretics, aspirin, and antihyperuricemic agents, could potentially contribute to hyperuricemia by either increasing uric acid production or impeding its excretion.

To determine the vital status of participants, we employed a probabilistic linkage method by associating NHANES personal identifiers with death certificates obtained from the National Death Index up until December 31, 2019. Specifically, the identification of incident cardio-cerebrovascular mortality was based on the underlying cause of death (I00–I09, I11, I13, I20–I51, and I60–I69), determined through the application of the International Classification of Diseases, Tenth Revision (ICD- 10) coding system.

Statistical analysis

To ensure the statistical robustness of our analyses, we implemented various adjustments to account for the complex sampling design of NHANES. We applied weighting techniques to account for the unequal probabilities of selection due to oversampling of certain subpopulations. This helps to generalize our findings to the overall population accurately. In addition, we used the “dietary day one sample weight” to correct for potential biases resulting from non-response or under-reporting of foods or nutrients, thereby improving the accuracy of data analysis. In order to assess the statistical significance of differences in characteristics across SUA categories, weighted percentages were employed to present the sample size for categorical variables, while the Chi-square test was conducted to evaluate the association between these categorical variables.

To examine the association between SUA and ASCVD in cross-sectional analyses, we employed multivariable logistic regression. We used three progressively adjusted models: Model 1 adjusted for age, gender, and race; Model 2 (full adjusted model) further adjusted for age, gender, race, education level, marital status, annual household income, smoking status, alcohol consumption, BMI, diabetes mellitus, hypertension, hyperlipidemia, chronic kidney disease, physical activity, total energy intake, HEI- 2015, and the use of certain medications affecting uric acid metabolism. We also explored the potential non-linear association between SUA and ASCVD using restricted cubic spline (RCS) models.

In prospective analyses, Kaplan–Meier survival analysis was conducted to compare the survival probabilities among SUA quartiles. A log-rank test was used to assess statistical differences between the survival curves. Then, we utilized a weighted Cox proportional hazards model to assess the relationship between SUA and the risk of all-cause and cardio-cerebrovascular mortality among ASCVD individuals. To mitigate the impact of reverse causality, we excluded individuals who died within 12 months of the baseline examination. Similar to the cross-sectional analyses, we employed three progressively adjusted models and RCS models with three knots. To assess potential gender differences in the association between SUA and mortality outcomes, we conducted a sensitivity analysis stratified by sex in ASCVD patients. This analysis specifically examined whether SUA had different impacts on all-cause mortality and cardiovascular mortality in men and women, considering the potential role of menopause in modifying SUA metabolism and cardiovascular risk.

All statistical analyses were conducted using the R software, version 4.2.2, which provide a wide range of statistical tools and functions. We considered statistical significance at a threshold of P values less than 0.05 (two-tailed), indicating a low probability of obtaining the observed results by chance alone.

Results

Cross-sectional study: the association between SUA and ASCVD

Supplemental Table 2 provided a summary of the baseline characteristics based on SUA grouping. 34,447 participants were ultimately included in this study, representing 151.6 million noninstitutionalized residents of the United States. The overall prevalence of ASCVD was 6.866%, and stroke risk increased with increasing SUA levels. A statistically significant correlation was obtained for almost all outcomes across the SUA grouping, except for cyclosporine use.

Table 1 displays the association between SUA and ASCVD risk among the overall participants. When SUA was included as a continuous variable, no statistically significant association was found between SUA concentrations and ASCVD risk in Model 2. On the other hand, when SUA was divided into four clinically relevant groups, individuals in Group 4 exhibited a 48.1% higher risk of ASCVD in Model 2, but the trend test results in model 2 were not statistically significant. Multivariable RCS analyses revealed a U-shaped trend between SUA and ASCVD risk in all the participants (non-linear P value = 0.014) (Fig. 2).

Table 1 Association of SUA and ASCVD

Fig. 2

The non-linear association between SUA and ASCVD

Prospective study: association of SUA with all-cause and cardio-cerebrovascular mortality in ASCVD individuals

In this study, 3,060 individuals, representing 10.4 million noninstitutionalized residents, were included after excluding 59 individuals who died within 1 year of follow-up to mitigate reverse causality. During a median follow-up period of 8.3 years, there were 1,233 deaths, including 524 cardio-cerebrovascular deaths. Supplemental Table 3 provided a summary of the baseline characteristics based on the grouping of SUA levels. The Kaplan–Meier survival curves demonstrate the association between SUA levels and survival probability (Supplemental Fig. 1). Participants in the second SUA quartile (Q2) had the highest survival probability, while those in the lowest (Q1), third (Q3), and highest (Q4) SUA quartiles exhibited significantly lower survival rates. Log-rank tests confirmed a statistically significant difference among the SUA groups (P < 0.0001).

The relationships between SUA and the risks of all-cause mortality are presented in Table 2. In the fully adjusted model, the hazard ratios (HRs) (95% CI) for all-cause mortality were 1.290 (1.085, 1.533), 1.00 (reference group), 1.451 (1.168, 1.803), and 2.132 (1.408, 3.226) for Group 1 to Group 4, respectively. Similarly, a comparable trend was observed for the association between SUA and cardio-cerebrovascular mortality, as shown in Table 3. The HRs (95% CI) for cardio-cerebrovascular mortality were 1.496 (1.070, 2.091), 1.00 (reference group), 1.664 (1.146, 2.416), and 2.700 (1.388, 5.250) for Group 1 to Group 4, respectively.

Table 2 Association of SUA and the risks of all-cause mortality among ASCVD individuals

Table 3 Association of SUA and the risks of cardio-cerebrovascular mortality among ASCVD individuals

These findings indirectly suggested that the relationship between SUA and mortality rates was not simply linear. To further explore this, we conducted RCS analyses. Notably, significant U-shaped trends were observed between SUA and the risks of all-cause mortality. The inflection point was estimated to be 5.748, with a non-linear p value < 0.0001 (Fig. 3). On the left side of the inflection point, SUA acted as a protective factor for all-cause mortality, while on the right side of the inflection point, SUA became a risk factor for all-cause mortality. Similarly, a similar U-shaped trend was observed for the association between SUA levels and cardio-cerebrovascular mortality, with an inflection point of 6.936 and a non-linear p value < 0.0001 (Fig. 4). In the sex-stratified analysis, we observed a U-shaped association between SUA levels and mortality outcomes in both men and women. Specifically, individuals with either low or high SUA levels exhibited increased risks of all-cause mortality and cardio-cerebrovascular mortality compared to those with moderate SUA levels (Supplemental Fig. 2).

Fig. 3

The non-linear association between SUA and all-cause mortality among ASCVD individuals

Fig. 4

The non-linear association between SUA and cardio-cerebrovascular mortality among ASCVD individuals

Discussion

Despite the ongoing debate surrounding the causality between SUA levels and ASCVD, there was a burgeoning interest in exploring this relationship given the rising prevalence of hyperuricemia on a global scale [22]. In our study, we employed a combination of cross-sectional analyses and prospective cohort study methodologies to investigate the correlation between SUA levels and the prevalence of ASCVD, as well as the risks of all-cause mortality and cardio-cerebrovascular mortality among ASCVD individuals. Our findings revealed a distinctive U-shaped pattern in the association between SUA and the risk of ASCVD among all participants. Similarly, a comparable trend was observed in the relationship between SUA levels and the incidence of all-cause mortality and cardio-cerebrovascular mortality among ASCVD individuals.

The correlation between elevated SUA levels and indicators of cardiovascular disease risk has been consistently demonstrated in numerous epidemiological and clinical studies. For instance, Kannel et al. conducted the Framingham study in 1967, involving a 12-year follow-up of 5,127 participants. The findings revealed that hyperuricemia was associated with an increased risk of coronary heart disease in men aged 30–59 years [23]. Similarly, the Brisighella Heart Study reported a significant association between elevated SUA levels and hypertension and atherosclerosis [24]. More recently, Zhang et al. conducted a cross-sectional study that identified a significant link between elevated SUA levels, traditional cardiovascular risk factors, and a 10-year ASCVD risk in residents of Xiamen [25]. Chen et al. found that hyperuricemia was associated with an increased 10-year risk of ASCVD in a Chinese adult population, and this association varied significantly between males and females [26]. Likewise, Tian et al. conducted a cross-sectional study in Jiangsu Province, China, which revealed variations in the prevalence of hyperuricemia across different survey sites. Cardiovascular risk factors were also found to increase with higher SUA quartiles, while HDL levels decreased. Cardiovascular risk factors also increased with increasing SUA quartiles. In addition, HDL levels decreased with increasing SUA quartiles. However, this study also observed gender differences in the association between SUA levels and 10-year ASCVD risk scores in China, with U-shaped associations found in men and J-shaped associations found in women [27]. Our findings are further supported by the meta-analysis conducted by Mazidi et al., which demonstrated a strong association between elevated SUA levels and increased risk of total and cause-specific mortality, emphasizing the importance of maintaining SUA within an optimal range [28]. In addition, the ATTICA study by Katsiki et al. highlighted the significant association between SUA levels and 10-year cardiovascular disease incidence, underscoring the role of SUA as an independent cardiovascular risk predictor [29]. The evidence presented in our study aligns with prior research, underscoring the correlation between elevated SUA levels and ASCVD risk indicators. Both low and high SUA levels were associated with an increased risk of ASCVD, emphasizing the potential of SUA as a biomarker for assessing ASCVD risk. However, in the fully adjusted model, the trend test results were not statistically significant. This can be attributed to the U-shaped relationship between SUA levels and the outcomes. In the early portion of the curve, SUA primarily exhibits protective effects, whereas in the later portion, SUA levels act as a risk factor. This pattern suggests that SUA has an optimal range, beyond which the risk of ASCVD increases. The non-significant trend test reflects the complexity of this non-linear association, which cannot be adequately captured by a simple linear trend test. To validate these findings, we conducted a prospective cohort study and observed a consistent U-shaped trend in the relationship between SUA levels and the incidence of both all-cause mortality and cardio-cerebrovascular mortality in individuals with ASCVD. These results suggest that deviations from the optimal range of SUA levels may confer an elevated risk of adverse cardio-cerebrovascular outcomes.

Elevated levels of SUA play a significant role in promoting oxidative stress and contribute to the pathogenesis of atherosclerosis. They induce dysfunction in endothelial cells, inflammation in macrophages, platelet aggregation, and oxidation of low-density lipoprotein, thereby exacerbating the progression of atherosclerosis [30]. Given the hypothesis that hyperuricemia is an independent risk factor for ASCVD, numerous studies have demonstrated that interventions aimed at lowering SUA levels can effectively reduce the incidence and progression of ASCVD. These interventions have served as valuable tools in exploring the specific causal relationship between SUA and ASCVD [31]. A systematic review and meta-analysis reported that treatment with xanthine oxidase inhibitors improved endothelial function and circulating markers of oxidative stress in patients at risk for CVD [32]. Similarly, a randomized, placebo-controlled trial demonstrated that high-dose allopurinol, a xanthine oxidase inhibitor, may reduce mortality in patients with CVD by mitigating vascular oxidative stress and improving endothelial dysfunction [33]. However, it is noteworthy that not all studies have shown a beneficial effect of uric acid reduction on cardiovascular disease. A recent meta-analysis of randomized controlled trials conducted in patients with gout revealed that urate-lowering therapy did not lead to a reduction in cardiovascular disease mortality compared to placebo. These findings suggest that aggressive SUA lowering may not always be beneficial. As SUA has antioxidant and endothelial functions, over-reduction might impair vascular homeostasis. A balanced approach to SUA modulation, rather than indiscriminate lowering, may be needed. Moreover, in long-term trials comparing febuxostat (a non-purine xanthine oxidase inhibitor) to allopurinol, no decrease in cardiovascular disease events over time was observed, and, in fact, more cardiovascular disease events occurred after treatment with febuxostat. This suggests a potential adverse effect of excessively low SUA levels [31, 34]. Low levels of SUA may contribute to atherosclerosis by reducing nitric oxide synthesis and increasing oxidative stress, ultimately leading to vascular endothelial dysfunction [35].

While elevated SUA levels are widely associated with cardiovascular risk, our study also found an increased risk at lower SUA levels, forming a U-shaped association. This paradoxical finding may be explained by non-traditional risk factors. Low SUA levels may reflect poor metabolic reserve, chronic disease burden, anemia, or impaired renal function, which independently increase cardiovascular mortality. SUA is also involved in erythropoiesis and endothelial function, and its deficiency may contribute to vascular dysfunction. These findings suggest SUA could serve as a marker of overall health status, rather than just a cardiovascular risk factor. SUA serves as a key antioxidant, neutralizing reactive oxygen species (ROS) and protecting endothelial function, and its deficiency may lead to oxidative stress and endothelial dysfunction. In addition, SUA plays a role in nitric oxide (NO) homeostasis, which is crucial for vascular function, and lower SUA levels have been linked to reduced NO bioavailability, impairing vasodilation and promoting vascular dysfunction. Moreover, lower SUA levels may indicate malnutrition or frailty, which are strongly associated with increased cardiovascular risk, particularly in the elderly or critically ill. These mechanisms support our findings that both low and high SUA levels are linked to increased cardiovascular events, emphasizing the need to maintain SUA within an optimal range.

Furthermore, the U-shaped association of SUA with all-cause and cause-specific mortality has been consistently observed across different populations. A prospective cohort study involving 9,118 US adults confirmed a U-shaped relationship between UA levels and mortality risk, with a SUA level of 5.7 mg/dL identified as the inflection point of the U-shaped curve [36]. Similarly, a cohort study of 375,163 adults in Korea demonstrated that both low and high SUA levels were significantly associated with elevated all-cause and CVD mortality [37]. Although the cutoff value we observed may slightly differ from other studies, variations in subjects, clinical characteristics, sample size, grouping strategies, and adjustment for confounding factors may account for these differences. However, the overall trend remains consistent, with the inflection point generally falling within the range of 5.5–6.5 mg/dL. These findings have the potential to inform clinical guidelines for the prevention and treatment of ASCVD.

Strengths and limitations

Our study had several merits. By virtue of its substantial sample size and long-term follow-up, it achieved strong statistical power, thereby mitigating selection bias and increasing the reliability of our findings. Moreover, we diligently adjusted for a diverse array of potential confounding factors, thereby bolstering the trustworthiness of our results. Nevertheless, our study also harbored certain limitations. Primarily, the information pertaining to ASCVD in NHANES was reliant on self-reported historical accounts, thus potentially engendering recall bias and subsequent misclassification of ASCVD cases. Second, while our study excluded several common medications known to significantly affect SUA levels, such as antihyperuricemic agents, other drugs like SGLT2 inhibitors may also influence SUA levels but could not be individually accounted for [38]. Third, the participants enrolled in our study were exclusively US civilians, necessitating further corroboration before extrapolating the outcomes to other populations. Fourth, the observational nature of our study precluded us from establishing a causal relationship between serum SUA levels and ASCVD or mortality risk.

Conclusions

Our findings elucidated a U-shaped association between SUA levels and the risks of ASCVD. Furthermore, we have also observed a similar U-shaped pattern between SUA levels and the risks of all-cause mortality and cardio-cerebrovascular mortality among individuals with ASCVD. Nonetheless, there remained a need for additional investigation to enhance our understanding of the underlying mechanisms through which SUA impacts ASCVD. In doing so, we aspired to uncover novel prevention strategies and guidelines aimed at alleviating the burden of ASCVD on human health.