Association between annual changes and visit-to-visit variability of serum uric acid and the kidney outcome in a general population

Abstract

Background

To determine whether variability of serum uric acid (UA) is associated with risk of chronic kidney disease (CKD) in a healthy population.

Methods

Retrospective, longitudinal cohort study was conducted at a health examination center in China. The study enrolled subjects who had a minimum of three visits between 2011 and 2018. We assessed UA change and visit-to-visit UA variability including standard deviation (SD), coefficient of variation (CV), variability independent of the mean (VIM), and average real variability (ARV). Rapid estimated glomerular filtration rate (eGFR) decline was defined by annual eGFR change < −4 mL/min/1.73 m²/year. We conducted a multivariable-adjusted logistic regression analysis.

Results

Ten thousand seven hundred and thirty-eight participants were included. During 4.43 ± 1.31 years follow-up, there were 535 cases with rapid eGFR decline and 240 cases developed CKD. Compared to the non-rapid eGFR decline group and non-CKD group, the UA annual changes and variability were higher in the rapid eGFR decline group and CKD group. The highest quartile of UA annual changes and variability showed a higher incident rate of rapid eGFR decline and that of CKD. After adjusting for covariates, OR for eGFR rapid decline in UA variability were 1.69 [1.53, 1.86] for annual changes of UA, 1.17 [1.08, 1.27] for SD of UA, 1.16 [1.06, 1.25] for CV of UA, 1.16 [1.07, 1.25] for VIM of UA, and 1.10 [1.02, 1.19] for ARV of UA. Consistent results were observed when CKD is used as the outcome.

Conclusions

Higher variability of serum UA was independently associated with the risk of kidney impairment.

Keywords:

• Chronic kidney disease
• uric acid
• variability
• annual uric acid changes
• risk factors

Introduction

The prevalence of chronic kidney disease (CKD) in China is experiencing a steady increase, with an incidence rate of 10.8–11.8% [1,2]. The management of end-stage kidney disease places a substantial burden on public healthcare systems. Hyperuricemia (HUA) is an independent risk factor for incident of CKD [2]. The relation between uric acid (UA) and CKD is mostly from studies focused on baseline serum UA levels and incident of CKD. Solely relying on a single baseline measurement of serum UA may not accurately represent the long-term exposure to serum UA and the associated risk of CKD. Moreover, a previous study has demonstrated a J-shaped association between serum UA levels and renal function [3]. Some studies have identified low serum UA as a risk factor for rapid decline of kidney function [4]. One retrospective study indicated that UA-lowering therapy (ULT) was not associated with favorable kidney outcomes in patients with normal baseline kidney function. Moreover, it may even pose potential risks for patients with moderately elevated serum UA levels [5]. So, whether the fluctuation of serum UA plays a role in the incident of CKD and how it influences the CKD is not sure.

Actually, fluctuations of indexes, such as glucose [6], body weight [7], always related to diseases. Recently, observations have demonstrated that increased variability of blood pressure [8], hemoglobin A1c [9], and metabolic factors [10] were associated with increased risk of CKD or kidney related outcomes. Studies found that visit-to-visit variability of UA associated with the risk of all-cause mortality in the general population [11]. It has been demonstrated that there was a correlation between fluctuations and changes in serum UA levels and the occurrence of gout flares with intensive ULT [12]. However, annual changes and the visit-to-visit variability of UA and the risk of CKD were not fully elucidated. So, the objective of this study is to examine the relationship between changes and visit-to-visit variability of serum UA levels and the occurrence of CKD in a general population.

Methods

Study design and covariates

The Tonglu First People’s Hospital Health Examination Center is one of the largest healthcare facilities in the Eastern Area of Zhejiang Province, specifically in Tonglu. A retrospective longitudinal study was conducted at this center from January 2011 to November 2018. The flowchart of the screening program is presented in Figure S1. Screened participants were identified using their ID number, birthdates, and other identifying information. Participants older than 18 years of age were included. The exclusion criteria for this study were as follows: (1) lacking serum creatinine (Scr) and UA measurements (n = 421); (2) baseline estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m² (n = 828); (3) fewer than three visits (n = 23,263); and (4) lacking relevant data, such as urine protein (n = 663). Thus, a total of 10,738 participants met the selection criteria for this study.

Venous blood samples were collected in the morning. An autoanalyzer (Roche, Indianapolis, IN) was used to measure all blood variables. The eGFR was calculated using the CKD Epidemiology Collaboration (CKD-EPI) [13]. HUA was defined by the sex-specific criteria by serum UA >7 mg/dL in males and >6 mg/dL in females [14]. Diabetes was defined as a fasting blood glucose (FBG) level of ≥7.0 mmol/L. According to the 2016 revised Chinese Adult Dyslipidemia Prevention Guide, dyslipidemia was defined by the presence of any of the following criteria: (1) serum total cholesterol (TC) ≥ 5.2 mmol/L; (2) triglyceride (TG) ≥ 1.7 mmol/L; (3) high-density lipoproteins (HDLs) < 1.0 mmol/L; (4) low-density lipoproteins (LDLs) ≥ 3.4 mmol/L. Otherwise, values within the normal range were classified as normal. Hyperlipidemia was defined as TC ≥ 5.20 mmol/L and TG ≥ 1.7 mmol/L.

Indices of serum UA variability include standard deviation (SD), the coefficient of variation (CV) [15], average real variability (ARV) [16], and variability independent of the mean (VIM) [17]. VIM is calculated as 100 × SD/mean to the power x, where x is the regression coefficient based on the SD in natural logarithm divided by the natural logarithm of the mean. ARV across multiple visits is determined by calculating the average absolute differences between successive UA measurements in each visit. SD is more related to the mean, VIM and CV reflect more variability. ARV is the average of the absolute differences between consecutive values and less influenced by the trend. We stratified participants into 4 categories according to quartiles of serum UA variability.

For each participant, annualized eGFR slope (mL/min/1.73 m²/year) and annualized UA slope (mg/dL per year) were derived from ordinary least squares regression [18] using all available eGFR and UA data. A rapid eGFR decline was defined by annual eGFR change < −4 mL/min/1.73 m²/year. CKD was defined by eGFR <60 mL/min/1.73 m² or decrease of eGFR ≥30%.

Statistical analysis

Descriptive statistics were employed to compare the characteristics of the cohorts based on CKD outcomes. The normal distribution of the data was assessed using the Kolmogorov–Smirnov test. Continuous variables were presented as means with their corresponding SDs. ANOVA was employed to compare normally distributed continuous variables. Bonferroni’s adjusted tests for significance were used to compare the groups. We conducted a multivariable-adjusted logistic regression analysis. Statistical analyses were conducted using SPSS 20 (SPSS Inc., Chicago, IL) and R software (version 3.44) (R Foundation for Statistical Computing, Vienna, Austria), and a two-sided p value of <.05 was considered significant.

Results

We included participants with at least three visits in the follow-up from January 2011 to November 2018 (Figure S1). Four thousand two hundred and fifteen were excluded as lack of UA and Scr data, 848 were excluded for baseline <60 mL/min/1.73 m², 23,263 were excluded for fewer than three visits, and 663 were excluded for lack of relevant data. We included 10,738 participants ultimately. During a mean of 4.43 years follow-up, 535 participants experienced rapid eGFR decline and 240 developed CKD. Then, we divided them based on CKD outcomes (Table 1, Figure S2 and Table S1). The annual UA changes and UA variability (SD, CV, VIM, and ARV) were higher in eGFR rapid decline group and CKD group. The baseline characteristics of participants are presented in Table 2 and S2-5 stratified by quartiles of serum UA variability. In Table 2, participants with greater SD of UA were older and more likely to be men. The albumin, total protein, hemoglobin (Hb), mean UA, FBG, cholesterol (CHOL), TGs, LDL-c, and BUN of these participants tend to be higher and baseline eGFR is lower. Incidence rates for rapid eGFR decline and CKD in Q4 of variability of UA were higher as expected (Table 3).

Table 1. Serum uric acid annual changes and variability in groups categorized by kidney outcomes.

	Non-rapid eGFR decline (N = 10,203)	Rapid eGFR decline (n = 535)	p Value	Non-CKD (N = 10,488)	CKD (N = 240)	p Value
Annual UA change (mg/dL/y)	−0.03 ± 0.37	0.19 ± 0.53	<.001	−0.02 ± 0.38	0.08 ± 0.55	.008
Mean UA (md/dL)	5.95 ± 1.37	6.55 ± 1.54	<.001	5.95 ± 1.36	7.19 ± 1.55	<.001
SD of UA (mg/dL)	0.89 ± 0.43	1.07 ± 0.62	<.001	0.90 ± 0.44	1.27 ± 0.68	<.001
CV of UA (%)	0.15 ± 0.07	0.16 ± 0.08	.002	0.15 ± 0.07	0.18 ± 0.08	<.001
VIM of UA	0.90 ± 0.39	0.99 ± 0.50	<.001	0.90 ± 0.39	1.10 ± 0.53	<.001
ARV of UA	1.08 ± 0.60	1.26 ± 0.79	<.001	1.08 ± 0.61	1.40 ± 0.80	<.001

Table 2. Baseline characteristics of each group categorized by serum uric acid SD quantiles.

SD of serum UA	Q1	Q2	Q3	Q4	p Value
N	2685	2684	2685	2684
Age (years)	47.43 ± 14.95	47.37 ± 14.67	48.65 ± 14.76	50.6 ± 14.82	<.001
Sex (male (%))	1193 (44.40)	1325 (49.37)	1488 (55.42)	1694 (66.11)	<.001
Albumin (g/L)	45.85 ± 2.42	45.80 ± 2.47	45.85 ± 2.44	45.77 ± 2.49	.552
Total protein (g/L)	75.75 ± 3.99	75.44 ± 3.91	75.4 ± 4.03	75.71 ± 4.05	.001
Hb (g/L)	138.38 ± 17.99	139.9 ± 18.11	140.92 ± 17.65	143.04 ± 17.29	<.001
FBG (mmol/L)	4.97 ± 1.16	4.93 ± 1.16	4.95 ± 1.22	5.09 ± 1.41	<.001
Cholesterol (mmol/L)	1.17 [0.81–1.76]	1.24 [0.85–1.81]	1.30[0.92–1.97]	1.51[1.03–2.28]	<.001
Triglycerides (mmol/L)	4.68 ± 1.13	4.72 ± 1.13	4.81 ± 1.13	4.97 ± 1.17	<.001
HDL-c (mmol/L)	1.34 ± 0.38	1.33 ± 0.37	1.33 ± 0.37	1.33 ± 0.36	.466
LDL-c (mmol/L)	1.98 ± 1.29	2.14 ± 1.25	2.15 ± 1.3	2.03 ± 1.41	<.001
UA (mg/dL)	5.35 ± 1.32	5.49 ± 1.38	5.71 ± 1.43	6.38 ± 1.81	<.001
Serum creatinine (μmol/L)	66.53 ± 13.94	68.05 ± 15.08	69.54 ± 15.15	71.01 ± 15.22	<.001
BUN (mmol/L)	5.22 ± 1.48	5.17 ± 1.39	5.24 ± 1.41	5.37 ± 1.39	<.001
eGFR (mL/min/1.73 m^2)	101.67 ± 14.14	101.24 ± 14.49	99.73 ± 14.33	97.96 ± 14.52	<.001
eGFR change (mL/min/1.73 m^2/year)	−0.63 ± 2.09	−0.70 ± 1.90	−0.72 ± 1.96	−0.93 ± 2.46	<.001
Serum UA change (mg/dL/year)	0.00 ± 0.17	−0.01 ± 0.26	−0.02 ± 0.35	−0.03 ± 0.61	.004
mean UA (md/dL)	5.43 ± 1.28	5.68 ± 1.24	6.01 ± 1.23	6.79 ± 1.37	<.001
SD of UA (mg/dL)	0.41 ± 0.14	0.73 ± 0.07	0.98 ± 0.08	1.49 ± 0.38	<.001
CV of UA (%)	0.08 ± 0.03	0.13 ± 0.03	0.17 ± 0.04	0.22 ± 0.06	<.001
VIM of UA	0.46 ± 0.18	0.78 ± 0.14	1.00 ± 0.17	1.37 ± 0.34	<.001
ARV of UA	0.52 ± 0.23	0.88 ± 0.24	1.17 ± 0.32	1.77 ± 0.68	<.001

Alb: albumin; BUN: blood urea nitrogen; Hb: hemoglobin; UA: uric acid; eGFR: estimated glomerular filtration rate; ARV: average real variability; CV: coefficient of variation; SD: standard deviation; VIM: variability independent of the mean.

Values for categorical variables are given as counts (%); values for continuous variables, as the mean ± standard deviation or median [interquartile range]. p Values represent differences across quartiles.

Table 3. Incidence rate of eGFR rapid decline or CKD with respect to quartiles of UA variability.

			eGFR rapid decline			CKD
	Quantiles		Events	Incidence rate	p Value	Events	Incidence rate	p Value
Annual UA change	Q1 (N = 2590)	−0.48 ± 0.30	95	3.50%	<.001	56	2.10%	<.001
	Q2 (N = 2620)	−0.11 ± 0.06	64	2.40%		45	1.70%
	Q3 (N = 2580)	0.09 ± 0.06	107	4.00%		45	1.70%
	Q4 (N = 2413)	0.44 ± 0.22	269	10.00%		94	3.50%
SD of UA	Q1 (N = 2571)	5.43 ± 1.28	114	4.20%	<.001	26	1.00%	<.001
	Q2 (N = 2579)	5.68 ± 1.24	105	3.90%		38	1.40%
	Q3 (N = 2565)	6.01 ± 1.23	120	4.50%		65	2.40%
	Q4 (N = 2488)	6.79 ± 1.37	196	7.30%		111	4.10%
CV of UA	Q1 (N = 2546)	0.07 ± 0.02	139	5.20%	.001	47	1.80%	<.001
	Q2 (N = 2571)	0.13 ± 0.01	113	4.20%		53	2.00%
	Q3 (N = 2570)	0.17 ± 0.01	115	4.30%		49	1.80%
	Q4 (N = 2516)	0.24 ± 0.05	168	6.30%		91	3.40%
VIM of UA	Q1 (N = 2551)	0.44 ± 0.15	139	5.20%	<.001	39	1.50%	<.001
	Q2 (N = 2584)	0.76 ± 0.07	113	4.20%		50	1.90%
	Q3 (N = 2560)	1.00 ± 0.07	115	4.30%		53	2.00%
	Q4 (N = 2508)	1.42 ± 0.30	168	6.30%		98	3.70%
ARV of UA	Q1 (N = 2570)	0.45 ± 0.15	115	4.30%	<.001	37	1.40%	<.001
	Q2 (N = 2571)	0.83 ± 0.09	114	4.20%		38	1.40%
	Q3 (N = 2563)	1.16 ± 0.11	121	4.50%		64	2.40%
	Q4 (N = 2499)	1.90 ± 0.57	185	6.90%		101	3.80%

In an unadjusted model (Table 4), each 1 − SD increase in annual UA changes (per 0.38 mg/dL; OR [95%CI] = 1.80 [1.64, 1.97]), SD of UA (per 0.45 mg/dL; OR [95%CI] = 1.38 [1.28, 1.48]), CV of UA (per 0.07; OR [95%CI] = 1.17 [1.07, 1.27]), VIM of UA (per 0.40; OR [95%CI] = 1.24 [1.14, 1.34]), and ARV of UA (per 0.61; OR [95%CI] = 1.28 [1.19, 1.38]) was associated with an increased risk for rapid eGFR decline. When sex, age, and mean UA were entered into the model that included variability of UA, the ORs [95%CI] for rapid eGFR decline for each 1 − SD increase for the annual UA changes were 1.60 [1.46, 1.76], 1.19 [1.10, 1.29] for the SD of UA, 1.17 [1.08, 1.27] for the CV of UA, 1.18 [1.09, 1.27] for the VIM of UA, and 1.12 [1.04, 1.21] for the ARV of UA. When BUN, FBG, Hb, total protein, albumin, TG, CHOL, HDL-c, and LDL-c were included, the ORs [95%CI] for rapid eGFR decline were 1.69 [1.53, 1.86] for annual UA changes, 1.17 [1.08, 1.27] for SD of UA, 1.16 [1.06, 1.25] for CR of UA, 1.16 [1.07, 1.25] for VIM of UA, and 1.10 [1.02, 1.19] for ARV of UA.

Table 4. ORs for study outcomes associated with 1 − SD increase in level of UA variability.

	UA change	p Value	SD	p Value	CV	p Value	VIM	p Value	ARV	p Value
1 − SD change	0.38		0.45		0.07		0.40		0.61
eGFR rapid change
Crude, OR [95%CI]	1.80 [1.64, 1.97]	<.001	1.38 [1.28, 1.48]	<.001	1.17 [1.07, 1.27]	<.001	1.24 [1.14, 1.34]	<.001	1.28 [1.19, 1.38]	<.001
Model, OR [95%CI]
1	1.60 [1.46, 1.76]	<.001	1.19 [1.10, 1.29]	<.001	1.17 [1.08, 1.27]	<.001	1.18 [1.09, 1.27]	<.001	1.12 [1.04, 1.21]	.004
2	1.69 [1.53, 1.86]	<.001	1.17 [1.08, 1.27]	<.001	1.16 [1.06, 1.25]	.001	1.16 [1.07, 1.25]	<.001	1.10 [1.02, 1.19]	.014
CKD
Crude, OR [95%CI]	1.30 [1.14, 1.48]	<.001	1.72 [1.58, 1.88]	<.001	1.37 [1.22, 1.53]	<.001	1.50 [1.35, 1.66]	<.001	1.44 [1.31, 1.57]	<.001
Model, OR [95%CI]
1	1.08 [0.95, 1.21]	.231	1.29 [1.16, 1.44]	<.001	1.33 [1.19, 1.50]	<.001	1.31 [1.18, 1.47]	<.001	1.12 [1.01, 1.25]	.031
2	1.28 [1.12, 1.47]	<.001	1.34 [1.20, 1.49]	<.001	1.37 [1.21, 1.54]	<.001	1.35 [1.20, 1.51]	<.001	1.15 [1.03, 1.28]	.014

UA: uric acid; CI: confidence interval.

Model 1 further adjusted for sex, age, and mean UA; model 2 further adjusted for BUN, FBG, hemoglobin, total protein, albumin, triglyceride, cholesterol, HDL-c, and LDL-c.

Also, each 1 − SD increase in annual UA changes (OR [95%CI] = 1.30 [1.14, 1.48]), SD of UA (OR [95%CI] = 1.72 [1.58, 1.88]), CV of UA (OR [95%CI] = 1.37 [1.22, 1.53]), VIM of UA (OR [95%CI] = 1.50 [1.35, 1.66]), and ARV of UA (OR [95%CI] = 1.44 [1.31, 1.57]) was associated with an increased risk for incidence of CKD. When sex, age, and mean UA were entered into the model that included variability of UA, the ORs [95%CI] for incidence of CKD for each 1 − SD increase for the annual UA changes were 1.08 [0.95, 1.21], 1.29 [1.16, 1.44] for the SD of UA, 1.33 [1.19, 1.50] for the CV of UA, 1.31 [1.18, 1.47] for the VIM of UA, and 1.12 [1.01, 1.25] for the ARV of UA. When BUN, FBG, Hb, total protein, albumin, TG, CHOL, HDL-c, and LDL-c were included, the ORs [95%CI] for incidence of CKD were 1.28 [1.12, 1.47] for annual UA changes, 1.34 [1.20, 1.49] for SD of UA, 1.37 [1.21, 1.54] for CR of UA, 1.35 [1.20, 1.51] for VIM of UA, and 1.15 [1.03, 1.28] for ARV of UA. So, the variability of UA was independent risk factor for the CKD. Subgroup analyses showed that participants with higher variability of serum UA tend to have higher risk of kidney end (Table S6). The sensitivity analyses were in the individuals with eGFR ≥ 70 mL/min/1.73 m² (Table S7).

Discussion

The main findings of this study revealed an association between annual changes in serum UA and UA variability with CKD. Those with higher variability of serum UA tend to have higher albumin, total protein, Hb, mean UA, FBG, CHOL, TG, LDL-c, BUN, and lower eGFR levels compared with stable serum UA. Moreover, these factors were found to be independent risk factors for CKD and rapid eGFR decline, even after adjusting for mean serum UA and other potential risk factors.

Serum UA has been widely acknowledged for its detrimental impact on various human diseases, including hypertension, diabetes, and other metabolic syndrome [19]. However, it is worth noting that serum UA also possesses potent antioxidant properties [20]. It is true that HUA is an independent predictor of CKD. There is also a direct relationship of UA with prevalence of CKD. Numerous observational studies have provided evidence on the correlation between serum UA levels and CKD, ESRD [21,22], as well as a more rapid decline in renal function [23]. However, certain studies have failed to establish a significant association between ULT and incident of CKD or kidney disease progression [5]. In our study, our data also showed that in the participants with the same baseline serum UA, the serum UA fluctuated differently during follow-up. Whether the fluctuation of serum UA is related to the occurrence of kidney disease is not very clear.

The present study demonstrated that both annual UA changes and UA variability were significantly associated with CKD and rapid eGFR decline. We use annual UA changes, SD of UA, CV of UA, VIM of UA, and ARV of UA to evaluate the fluctuation of UA. These indexes are mostly used to assess variation [11].

Studies have found an association between UA variability and diseases. Increased variability in serum UA levels was found to be independently associated with an elevated risk of all-cause mortality in the general population [11]. A study conducted in Israel, involving 10,059 male tenured civil servants and municipal employees, demonstrated that increased variability in serum UA measurements was predictive of both coronary heart disease and all-cause mortality [24]. In patients with coronary artery disease, a high variability in serum UA levels was associated with an elevated risk of experiencing adverse outcomes following percutaneous coronary intervention [25]. Previous studies have also found there might be a relation between UA changes and kidney outcomes. In type 2 diabetes, the variability of serum UA influenced the development of diabetic kidney disease [26]. However, a study showed that there was no significant correlation of serum UA fluctuation during follow-up and renal function in 1009 gout patients [27]. Iseki et al. investigated individuals who had participated in visits both in 1993 and 2003. They discovered that the increase in serum UA levels from 1993 to 2003 was a more potent independent risk factor for a decline in eGFR compared to the baseline UA, hypertension, or diabetes [28]. Rathmann et al. studied relatively young adults from the Coronary Artery Risk Development in Young Adults Study, and they found that changes in Scr were strongly positively related to increasing UA values [29]. In another retrospective study, they found that elevated baseline serum UA and increasing serum UA over time were independent risk factors for rapid eGFR decline over five years [30]. In our study, we evaluated the serum variability of the serum UA by a visit-to-visit method which allowed us to detect the relation between the variability of serum UA and CKD from a different perspective. We verified these indicators in the sub-group and found some discrepancies. This may be explained by the fact that the ARV index includes the temporality of UA, and many of the female participants experiencing significant UA changes were near menopause, a period when UA fluctuates greatly, thus introducing bias. SD is significantly influenced by the mean of UA. Since higher serum UA levels are more related to older age and male gender, the relationship between SD changes and renal endpoints is less significant in older male patients. This discrepancy may also be explained by the small sample size of patients over the age of 64. Our study indicated that annual changes of UA and UA variability might be important factor for kidney outcomes.

Actually, it is not clear why the large variation of UA is closely related to CKD. It is known that there is a J-shaped relationship between UA and all-cause mortality in non-CKD [31] and HD patients [32]. Also, UA had a J-shaped association with ESRD in patients with IgA nephropathy (baseline eGFR of 78 ± 31 mL/min/1.73 m²), especially in women [33]. Those results indicated that individuals with serum UA levels at either extreme are at higher risk of adverse events. Second, observational studies have shown that fluctuations in UA levels may aggravate the inflammatory process of gout. A large change of serum UA will increase the crystallization rate of urate, thereby stimulating the immune response and inflammatory response [34]. This may indicate that large fluctuations in serum UA levels may be harmful. Besides, patients with large serum UA variability tend to have more underlying diseases and poorer controlling of CKD risk factors. The above are only speculative reasons, and more mechanism research is needed.

The strengths of our study include a relatively large population and the utilization of multiple measurements of serum UA. However, our study has several limitations. The participants were solely from a health examination center, which resulted in a lack of data regarding the utilization of angiotensin receptor blockers and ULT, dietary intake, disease conditions. Since we only included participants with eGFR ≥60 mL/min/1.73 m², we were unable to draw conclusions about participants with eGFR <60 mL/min/1.73 m². In addition, this study is an observational design, so it is impossible to infer a causal relationship between the UA variability and CKD. Even after multivariable adjustments, residual confounding is always a problem. There may also be limitation caused by selection bias and data-driven methods.

Conclusions

This study suggested that higher variability in serum UA is associated with an increased risk of incidence of CKD independent of mean serum UA. Our research emphasizes the importance of maintaining long-term stable serum UA levels.

Author contributions

Research idea and study design: JL and WW; data acquisition: GY; data analysis/interpretation and statistical analysis: JL and YM. The first draft of the manuscript was written by JL and YM. The manuscript was critically revised by WW. All authors have read and approved the final manuscript.

Ethics approval

This study protocol (2016-1-10) was approved by the Institutional Review Board of the Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. It was in accordance with the principles of the Helsinki Declaration II.

Consent form

Informed consent was waived by the ethics committee as all the data were collected after de-identification. The data were only accessible to the principal investigator, ensuring confidentiality. The information was kept confidential throughout the entire study.

Acknowledgements

We thank all participants who were involved in this study. We are also grateful to the Tonglu First People’s Hospital health examination center for providing data for analysis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets used and/or analyzed during the current study are available from the correspondence author on reasonable request.

Additional information

Funding

This study is supported by grants from National Natural Science Foundation of China (81700647, 81870492), the National Key Research and Development Program (2016YFC1305402), and Key Projects of National Basic Research Program of China 973 (2012CB517700).