Introduction

Hypertension is a chronic disease characterized by elevated arterial blood pressure. Long-term hypertension is an important risk factor for cardiovascular diseases such as coronary artery disease and stroke [1, 2]. The number of people with hypertension has increased substantially due to population growth and aging [3]. Therefore, it is very important to optimize the hypertension risk stratification method to identify people at high risk of cardiovascular diseases and other complications.

Serum uric acid (SUA), the end-product of purine metabolism in humans, is excreted largely by the kidneys [4]. A large number of studies have shown that elevated SUA is a risk factor for hypertension [5, 6]. For example, a population-based cohort study in the United States showed that increasing quartiles of SUA were associated with the 10-year incidence of hypertension independent of smoking, alcohol intake and baseline kidney function [7]. Experimental studies have suggested that SUA may activate the renin angiotensin system and inhibit the release of endothelial nitric oxide, which is followed by a reduction in renal blood flow and elevation of systemic blood pressure [8].

In addition, many studies have reported that some metabolic indicators, such as body mass index (BMI) [9, 10], waist-hip ratio (WHR) [11], blood lipids [12, 13], fasting blood glucose (GLU) [12, 14], white blood cell [15], and alanine aminotransferase (ALT) [16], are significantly associated with hypertension and are closely related to the occurrence and development of hypertension. Moreover, these metabolic indicators have been found to be significantly correlated with SUA [17,18,19]. There may not be a direct correlation between SUA and hypertension, and their correlation may be partially or fully mediated by metabolic factors. For instance, the African-PREDICT study pointed out that the association between SUA and blood pressure is significant in young healthy women with increased adiposity but not in lean women or men [20]. A retrospective cohort study conducted in Japan by Kuwabara et al. reported that SUA is an independent risk factor for the onset of hypertension in lean subjects, and the risk of hypertension is significantly higher in lean subjects with metabolic syndrome than in lean subjects without metabolic syndrome [21].

However, the temporal relationship between SUA and these metabolic factors has not been elucidated by research. In this context, cross-lagged panel analysis has attracted extensive attention. It is an analytical strategy used to describe reciprocal relationships or directional influences between variables over time [22]. Mediation analysis could elucidate the association between two variables and quantify the effect of mediating variables [23]. Therefore, the purpose of our study was to revisit the association between elevated SUA concentrations and hypertension in our cohort, and then we further applied cross-lagged panel analysis and mediation analysis to explore the roles of relevant metabolic factors in the association of SUA with hypertension.

Methods

Study population

The present study was based on a functional community cohort established in 201024. The cohort was composed of 8671 current employees or retired employees (aged 30–65) of governments, schools, hospitals, factories, business, and service institutions in Xicheng District, representing most of the occupational population in urban Beijing. Those with somatic or psychiatric abnormalities were excluded at enrollment [24]. All the participants in the cohort underwent an annual physical examination at the health management center of Beijing Xuanwu Hospital, Capital Medical University. We collected relevant baseline data in 2015 and followed up the participants for 4 years. Each participant signed an informed consent form, and this study was approved by the Ethics Committee of Capital Medical University. Participants were excluded if they had hypertension, acute and chronic infectious diseases and severe liver and kidney diseases at baseline. Finally, 5330 subjects with normal blood pressure at baseline were included in our analyses (Fig. 1).

Fig. 1
figure 1

Flow diagram of participant recruitment

Full size image

Data collection and clinical measurements

Data on general information such as age, sex, current smoking, alcohol consumption and physical exercise were collected by standard questionnaires. Current smoking was considered smoking more than one cigarette per day. Alcohol consumption was defined as the intake of wine/beer/cider/spirits 1 time per week or more. Physical exercise was defined as walking or bike riding more than 15 min daily, doing sports or physical exercise more than 2 h weekly, or lifting or carrying heavy objects at work daily [25].

Weight and height were measured while the subjects were wearing light clothing and no shoes. BMI was computed as weight in kilograms divided by the square of height in meters. Waist circumstance (WC) and hip circumference were measured with a soft ruler while the subjects were standing upright, with their arms drooping naturally and their abdomen in a relaxed state. The waist and hip measurements were accurate to 0.01 cm. The WHR was equal to WC divided by hip circumference. Blood pressure was measured uniformly with a calibrated mercury sphygmomanometer. The participants rested in a quiet state for 5 min. The right elbow was exposed with the participant in a sitting position and placed at the same level as the heart. The space between the lower margin of the cuff and the space in front of the elbow was 2~3 cm, and systolic blood pressure (SBP) and diastolic blood pressure (DBP) were recorded. The average of the two blood pressure measurements was taken, and the result was accurate to 1 mmHg.

Biochemical index detection

Blood samples were collected in the morning from 7:00 to 8:30 after participants fasted for 1 night. After the subjects provided informed consent, 5 ml peripheral venous blood was collected from the cubital vein using EDTA (ethylene diamine tetra-acetic acid) for biochemical marker detection. A fully automatic biochemical analyzer (Hitachi, 7060) was used to detect  GLU, total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinine (Cr), white blood cell count (WBC) and serum uric acid (SUA).

Definition of hypertension

Participants with SBP above 140 mmHg, DBP above 90 mmHg, or self-reported use of antihypertensive agents were diagnosed with hypertension according to the JNC-7 criteria [26].

Statistical analysis

We summarized subjects’ demographic and biochemical characteristics using descriptive statistics, reporting the mean and standard deviation (SD) for normally distributed data or median and interquartile ranges for nonnormally distributed continuous variables stratified by sex. A correlation bubble diagram was drawn to assess the correlations between the variables in this study. To determine seasonal differences, we used the following definitions of seasons: winter is December, January and February; spring is March, April and May; summer is June, July and August; and autumn is September, October and November. The seasonal differences in SUA at baseline were compared using the Kruskal–Wallis H test.

Restricted cubic splines (RCS) were drawn to understand the relationship between uric acid and hypertension in the population, adjusting for age, sex, BMI, WHR, current smoking, alcohol consumption and physical exercise. Multivariable logistic regression models were used to analyze the association between hypertension and baseline SUA. To further explore the association between the change in SUA and hypertension, latent class trajectory models (LCTMs) were used to identify trajectories of SUA from 2015 to 2019 [27]. This is a specialized form of finite mixture modeling. Our models used second-order polynomials. The best-fitting number of trajectories was based on the lowest value of Bayesian information criterion (BIC). According to the slope of each trajectory, we assigned labels to the trajectories to facilitate interpretability. Then, receiver operating characteristic (ROC) curves were generated to estimate the effects of SUA on hypertension [28].

The cross-lagged panel analysis was used to measure the time relationship between SUA and relevant metabolic factors (BMI, WHR, GLU, TC, TG, LDL, HDL, Cr, AST, ALT, WBC). Cross-lagged panel analysis is a form of path analysis [29, 30] that also examines the reciprocal, longitudinal relationships between a set of interrelated variables [31].

Once the temporal relationship between metabolic factors and SUA was established, mediation models were constructed to examine whether these selected factors played a mediating role in the association between SUA and hypertension in men and women. In this model, X was SUA (predictor), M was the mediator, and hypertension was the outcome; mediators were determined by the cross-lagged path that was changed after SUA. This statistical approach has been applied successfully in previous studies to demonstrate the role of mediators [32,33,34,35]. The path coefficients were the beta (β) coefficients of the multivariable regression models and represented the magnitude and direction of associations between variables included in the model. We included age as a covariate to adjust for the mediating effect of these indicators on the association between SUA level and hypertension. R 3.6.3, SPSS 24.0 and Mplus Version 7.3 were used for the statistical analyses. All statistical tests were two-sided, with P < 0.05 considered to indicate statistical significance.

Results

Baseline characteristics

A total of 5330 participants were included in the final analysis. The baseline characteristics of participants according to quintiles of SUA stratified by sex are shown in Table 1. The correlation between variables at baseline in our study is shown in Supplementary Fig. 1. There was a correlation between SUA and SBP (r = 0.30) and between SUA and DBP (r = 0.30). The seasonal difference in SUA is shown in Fig. 2a. SUA was significantly higher in autumn than in other seasons.

Table 1 Baseline characteristics of normal blood pressure participants in the cohort according to quartiles of SUA stratified by gender
Full size table
Fig. 2
figure 2

Seasonal distribution of serum uric acid and trajectories of serum uric acid from 2015 to 2019. a seasonal distribution of serum uric acid; b trajectories of serum uric acid from 2015 to 2019; class 1: decreasing slowly and then increasing slowly at high concentration; class 2: maintaining a low concentration; class 3: maintaining a high concentration; class 4: rapidly rising and then falling at high concentration; and class 5: maintaining a moderate concentration

Full size image

Cumulative incidence of hypertension

In 2019, the cumulative incidence of hypertension in the general population was 10.9% (n = 580), and there was a higher cumulative incidence of hypertension in men (12.2% (n = 299)) than in women (9.8% (n = 281)). As the SUA concentration increased, the incidence of hypertension gradually increased in men and women (P < 0.001) (Supplementary Figure 2).

The association between SUA and hypertension

In the total population, we did not observe a nonlinear relationship between baseline SUA and incident hypertension after adjustment for age, sex, BMI, WHR, current smoking, alcohol consumption and physical exercise (P > 0.05) (Fig. 3a). In the sex-stratified analysis, we also did not identify nonlinear relationships between baseline SUA and incident hypertension after adjustment for age, BMI, WHR, current smoking, alcohol consumption and physical exercise (Fig. 3b, c).

Fig. 3
figure 3

Restricted cubic splines (RCS) and RRs between baseline SUA quartiles and hypertension in the study of the total population and stratified by sex. a RCS of total population; b RCS of men; c RCS of women; d RRs of total population; e RRs of men; f RRs of women. Model 1: adjusted for age and sex; Model 2: further adjusted for BMI, WHR, smoking, alcohol consumption, and physical activity; Model 3: further adjusted for SBP, DBP, GLU, TG, TC, HDL, LDL, Cr, AST, ALT, and WBC at baseline

Full size image

The prospective association between baseline SUA concentration and the risk of developing hypertension in the longitudinal study is shown in Fig. 3d. In multivariable logistic regression analysis, SUA concentration quartiles were correlated with incident hypertension. After adjusting for age and sex (Model 1), using the lowest quartile as a reference, the relative risks (RRs) (95% CI) were 1.48 (1.11–1.97), 2.03 (1.53–2.71), and 2.41 (1.77–3.27) for quartiles 2–4, respectively. In Model 2, after additional adjustment for BMI, WHR, smoking, alcohol consumption, and physical exercise, the RRs (95% CI) were 1.34 (0.99–1.79), 1.70 (1.27–2.28), and 1.81 (1.32–2.48) for quartiles 2–4, respectively. Finally, after further adjusting for SBP, DBP, GLU, TG, TC, HDL, LDL, Cr, AST, ALT, and WBC at baseline in Model 3, the RRs (95% CI) were 1.20 (0.88–1.63), 1.50 (1.10–2.05), and 1.57 (1.11–2.22) for quartiles 2–4, respectively.

In the sex subgroup analysis, as Fig. 3e shows, the highest quartile of SUA compared with the lowest quartile was significantly associated with a 57% increased risk of hypertension (RR = 1.57 [1.07–2.33]) for men in Model 3. However, in women (Fig. 3f), we did not find a statistically significant association between SUA and hypertension (P = 0.092) in Model 3, which indicated elevated SUA in men had a higher risk of developing hypertension.

We identified five distinct trajectories of SUA in 5330 individuals (Fig. 2b), characterized by decreasing slowly and then increasing slowly at high concentrations (class 1; 187 [3.5%]); maintaining a low concentration (class 2; 1630 [30.6%]); maintaining a high concentration (class 3; 1497 [28.1%]); rapidly rising and then falling at high concentrations (class 4; 29 [0.5%]); and maintaining a moderate concentration (class 5; 1987 [37.7%]). Using the class 5 trajectory as the reference trajectory, the result of multiple logistic regression analysis showed that “decreasing slowly and then increasing slowly at high concentration” and “maintaining a high concentration” trajectories were associated with a higher risk of hypertension (Supplementary Table 1).

Predictive efficacy of serum uric acid for hypertension

According to the results of multivariable logistic regression analysis, SUA was an independent risk factor for hypertension in Model 3 after adjusting for all relevant variables in our study. Based on this, we take Model 3 as the prediction model. To assess the predictive efficacy of the prediction model, we assessed the ROC curve, for which the AUC was 0.725, and the 95% CI was 0.708–0.741 (Supplementary Table 2). In the analysis of male and female stratification, the AUC of women (0.769) was greater than that of men (0.674), indicating that the prediction model was even more effective in women. Supplementary Figure 3 shows the ROC curves for the prediction in the total population and stratified by sex.

Cross-lagged panel analysis of the association between SUA and BMI, WHR, Cr, GLU, TC, TG, HDL, LDL, ALT, AST, and WBC

Figure 4 shows the cross-lagged path analysis between BMI, WHR, Cr, GLU, TC, TG, HDL, LDL, ALT, AST, WBC and SUA in the total population, adjusted for age, sex and follow-up days. We can see that the changes in Cr, TG, LDL, ALT, AST and WBC occurred after the change in SUA, so we chose these indicators as potential mediators to analyze the association between SUA and hypertension. For these six indicators, both of the path coefficients ρ1 and ρ2 were significantly different from 0 (P < 0.001). ρ1 from baseline SUA to follow-up Cr (TG, LDL, ALT, AST and WBC) was greater than ρ2 from baseline Cr (TG, LDL, ALT, AST and WBC) to follow-up SUA (P < 0.001). The χ2 test for model fit showed a difference between the hypothesized model and the observed data (P < 0.001), which was probably related to the high sensitivity of the large sample size.

Fig. 4
figure 4

Cross-lagged path analysis of SUA and BMI, WHR, GLU, HDL, LDL, TC, TGs, Cr, WBC, ALT and AST adjusted for age, sex and follow-up days. a cross-lagged path of SUA and BMI, b cross-lagged path of SUA and WHR, c cross-lagged path of SUA and GLU, d cross-lagged path of SUA and HDL, e cross-lagged path of SUA and LDL, f cross-lagged path of SUA and TC, g cross-lagged path of SUA and TG, h cross-lagged path of SUA and Cr, i cross-lagged path of SUA and WBC, j cross-lagged path of SUA and ALT, k cross-lagged path of SUA and AST. r1: synchronous correlations; r2 and r3: tracking correlations; ρ1, ρ2: cross-lagged path coefficients. *: represent coefficients different from 0, P < 0.001. #: P < 0.001 for difference between ρ1 and ρ2

Full size image

Mediation analysis between SUA and hypertension

We tested Cr, TG, LDL, ALT, AST and WBC as potential mediators of the association between SUA and hypertension in men and women, with adjustment for age (Fig. 5 and Supplementary Table 3). TG, WBC and ALT played an intermediary role in both men and women, and the proportion of mediation in men was greater than that in women (TG: 14.76% > 14.55%; WBC: 11.61% > 8.55%; ALT: 15.93% > 6.89%). However, LDL and Cr did not play a mediating role in the relationship between SUA and hypertension in either men or women. Moreover, AST played a mediating role in men (AST: 11.17%) but not in women.

Fig. 5
figure 5

Mediation analysis of the association between SUA and hypertension. a men; b women

Full size image

Discussion

In our study, we found a positive association between elevated SUA concentration and hypertension, and the hypertension prediction model had a good predictive ability. Through the cross-lagged panel analysis, we distinguished six indicators that changed after the SUA level changed, and then we analyzed their mediating effects on the association between SUA and hypertension. According to the results of mediation analysis, we found that TG, WBC and ALT played important mediating roles in the association between SUA and hypertension both in men and women.

Consistent with the Chinese Hypertension Guidelines [36], the incidence of hypertension was significantly higher in males than in females. Consistent with other studies, our study showed that elevated SUA was a risk factor for the incidence of hypertension. In a retrospective, single-center cohort study in Japan, increased SUA was a strong risk factor for developing hypertension in individuals with prehypertension [5]. Increasing levels of SUA were associated with elevated blood pressure in a cohort study in the US population [37]. Some studies in the US and Japan have shown that uric acid is associated with a U-shaped curve in the development of cardiovascular disease or cardiovascular mortality and that low levels of uric acid have some harmful effects on the cardiovascular system [38,39,40]. Unfortunately, low SUA levels did not show a deleterious effect on the development of hypertension in our study. This may be related to differences in the age, ethnicity, region and diet of the subjects in the different studies. Therefore, whether low SUA levels in the Chinese population have harmful effects on hypertension and the cardiovascular system needs to be further explored in other national studies.

Several pathogenetic mechanisms have been implicated in the association between SUA and hypertension. High SUA may lead to renal vasoconstriction through inhibition of the nitric oxide pathway and through activation of the renin–angiotensin system [41,42,43] and then to increased blood pressure. Moreover, SUA may cause oxidative stress by stimulating the activity of NADPH oxidase. This triggers activation of immune cells in the kidney, which results in renal vasoconstriction and increased blood pressure [44]. Uric acid can be produced by xanthine oxidase (XO) catalyzing the oxidation of hypoxanthine and xanthine [45]. Kario et al. found that topiroxostat, an XO inhibitor, significantly reduced plasma XO activity, resulting in a reduction in blood pressure and urinary albumin [46]. This finding also demonstrated that uric acid could have an effect on blood pressure. Therefore, in clinical practice, treatment can be used to reduce the activity of xanthine oxidase, thereby reducing uric acid levels and the influence of uric acid on blood pressure, ultimately reducing the risk of hypertension. To further explain the association between SUA and hypertension, we evaluated the mediation effect of the metabolic factors whose changes were induced by SUA changes. It was noteworthy that TG played the largest mediating role in this association, suggesting that TG is critical in the pathogenesis of hypertension. In a population-based cohort study, mediation analysis also showed that TG played an important mediating role in the pathway linking SUA and hypertension [47]. Unlike our study, that study also found that TC, HDL, BMI, and GLU had a mediating effect on the association. Renin–angiotensin system activation was induced not only by triglycerides but also by uric acid concentration, and the two factors had joint effects on blood pressure [47, 48]. Experimental studies have shown that SUA could increase TG accumulation [49, 50] and that hyperuricemia could also increase TG levels [51]. In general, many studies have suggested that SUA is independently related to TG [52, 53], and both are risk factors for hypertension. However, the chronological or causal relationship between them is not yet clear. The results of our prospective study showed that the elevation of SUA caused the elevation of TG. The important mediating role of TG also illustrated this point. Therefore, in the process of hypertension prevention and treatment in the future, attention should be given to whether there are abnormal TG levels.

WBC is a marker of subclinical or low-grade inflammation, and its five subtypes play various roles in the inflammatory response and host immunity [54, 55]. Accumulating evidence suggests that inflammation contributes to the development and progression of hypertension [56]. Our study found that WBC played a certain mediating role in the association between SUA and hypertension, suggesting that SUA may lead to increased blood pressure by inducing inflammation. Recent studies have found that SUA may cause inflammation by activating both mitogen-activated protein kinases (MAPKs) as well as stimulating NADPH oxidase [44]. In this study, we also found that the elevation of both AST and ALT occurred after the elevation of SUA. ALT played a mediating role in both genders, but AST only had a mediating effect in men. This difference may be related to age differences between men and women and differences in liver function due to men drinking more alcohol. Both ALT and AST are markers of liver dysfunction, and studies have shown that ALT is strongly correlated with the onset of hypertension in the Chinese population [57, 58]. In addition, studies have shown that serum ALT is closely related to insulin resistance [59], excessive inflammation and oxidative stress [60]. Thus, SUA may affect liver function through certain pathways, resulting in abnormal liver metabolism and elevated AST, which in turn affects the occurrence and development of hypertension. However, there are currently no studies showing a significant association between AST and hypertension [61]; therefore, the mediating effect of AST needs to be further studied.

In addition, it is worth noting that the mediating proportions of all the metabolic factors that play an intermediary role in this study are greater in males than in females, suggesting that the pathogenesis of hypertension in males and females may be different due to sex, lifestyle habits, sex hormones [62] and physical activity, which needs further research. Additionally, sex stratification should be carried out in the future for the prevention and control of hypertension.

Our study had several strengths. First, this study was a 4-year longitudinal study, which allowed for better verification of the causal relationship between SUA and hypertension. Second, we added the cross-lagged panel analysis to analyze the time sequence relationship between the variables in the study. This study had some limitations. First, we did not quantitatively investigate alcohol intake, and there was a lack of lifestyle variables such as dietary habits [63], which may affect the relationship between uric acid and hypertension. Second, the fact that our study is a single-center analysis and the included subjects may be more concerned about health issues than the general population may lead to selection bias. Third, most of the population in this study was composed of residents in the Beijing area, which resulted in certain regional limitations and limited the generalizability of the research results.

Perspective of Asia

Although uric acid is an independent risk factor for hypertension that has been confirmed by many studies, it has been reported that there are some differences in the strength of the correlation between different races [64]. Therefore, further studies are needed to confirm the roles of metabolic factors (TG, WBC, ALT) in mediating the association between uric acid and hypertension in other ethnic groups.

Conclusions

In conclusion, elevated SUA concentration was an independent risk factor for hypertension in the Chinese population. Moreover, TG, WBC and ALT played important mediating effects on this association. The findings of our cross-lagged panel analysis and mediation effect analysis of SUA and hypertension development may further strengthen our understanding of the pathogenesis of hypertension, emphasizing the combined effect of serum uric acid and metabolic factors and providing a new direction and thinking for the prevention of hypertension.