Remnant cholesterol elevates hyperuricemia risk in the middle aged and elderly Chinese: a longitudinal study

Abstract

Evidence regarding the relationship between remnant cholesterol (RC) and hyperuricemia is limited. The purpose of this study is to investigate the association between RC and hyperuricemia in the middle aged and elderly Chinese. Information was extracted from the China Health and Retirement Longitudinal Study (CHARLS) survey 2011 and survey 2015. Four logistic regression models were established. Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) were applied to balance the baseline. Next, sensitivity analyses and restricted cubic spline (RCS) analysis were conducted to further explore the association. Cross-lagged panel model (CLPM) and mediation analysis were used to deduce the causal relationship between RC and hyperuricemia. This study contained 6,447 participants. A positive association between high RC and hyperuricemia was found in the full adjusted model (OR:1.80, P < 0.001). Similar results were also seen after PSM (OR:1.86, P < 0.001), IPTW (OR:1.80, P < 0.001) and sensitive analysis in non-overweight subgroups (OR:1.77, P < 0.001). Though non-linear relationship was not observed, CLPM exhibited that high level of RC can directly cause increase of blood uric acid (standardized β = 0.005, P < 0.001). Moreover, mediation analysis suggested that the positive association can be mediated by hypertension (β = 0.024; p = 0.004), CRP (β = 0.050; p < 0.001) and WBC (β = 0.024; p = 0.010). High level of RC is an independent risk factor for hyperuricemia, which can be mediated by inflammation and hypertension.

Introduction

Hyperuricemia is defined as an abnormal elevated status of serum uric acid, which is related to purines metabolic disorders¹. In China, hyperuricemia has become the second metabolic disease only to diabetic mellitus (DM), and its’ incidence presents an escalating trend by years².

Though most hyperuricemia are asymptomatic³, it could greatly impair the health status. The main manifestation of hyperuricemia is gout and gouty arthritis, which can cause terrible pain and constrained movement⁴. Besides, elevated uric acid is linked with other disorders such as atherosclerosis, cardiovascular disease, chronic renal disease, etc.^5–10. Till now, many risk factors for hyperuricemia have been testified. For example, hypertension and obesity have been suggested to induce hyperuricemia^11,12. Inflammation also can interact with uric acid and promote hyperuricemia^13,14. Further, lipid accumulation is proved to increase the hyperuricemia risk¹⁵. However, the association between RC and hyperuricemia has rarely been researched.

RC is the cholesterol content of triglyceride (TG)-rich lipoproteins¹⁶. It consists of very low-density lipoproteins (VLDL), intermediate-density lipoproteins (IDL), and chylomicron remnants¹⁷. In recent years, RC has been highlighted in the lipid management for its’ features. It has been demonstrated that RC but not low-density lipoprotein cholesterol (LDL-C) was the independent risk factor for incident cardiovascular disease in overweight subjects¹⁸. Similarly, elevated RC is associated with a higher risk of ischemic stroke and atherosclerosis^19,20. In addition, RC can affect the progress of DM. Evidence has been found that RC also associated with DM by mediating insulin resistance¹⁶, and its’ concentration can predict the progression of diabetic nephropathy and severe diabetic retinopathy in type 1 diabetes²¹. But to the present, association between RC and hyperuricemia has not received much attention.

Therefore, this study aimed to explore whether RC could induce hyperuricemia and tried to explain the way of RC affecting hyperuricemia. To our knowledge, this is the first longitudinal study on the relationship of RC and hyperuricemia in Chinese. The findings of this paper may provide a new basis for predicting and treating hyperuricemia.

Methods

Data source and study population

In this study, data was extracted from CHARLS. CHARLS is a large-scale public database designed to investigate the middle-aged and older adults in China. It covers 450 communities and villages in 150 counties and districts of 28 provincial administrative regions of the country²². Healthy status, physical measurements, basic personal information, family structure and economic support, etc. were contained in the surveys. The baseline survey was wave 2011, and was tracked biennially. CHARLS was approved by the Biomedical Ethics Committee of Peking University and informed consent was obtained from all participants. The permission of accessing CHARLS database had been acquired, and all procedures in this study were performed adheres to the Declaration of Helsinki.

Data used in this study was from survey 2011 and 2015. 17,708 participants were involved at baseline. The exclusion criteria for participants were as follows: (1) aged less than 45 years old; (2) lost to follow-up in 2015; (3) without the information of RC in survey 2011 and hyperuricemia in 2015; (3) diagnosed with hyperuricemia in survey 2011. The study eventually comprised 6,447 participants altogether. Then, they were divided into two groups according the median of RC at wave 2011, and were described as quartile 1 (Q1) and quartile 2 (Q2). Figure 1 shows the flow chart for study participants.

Fig. 1.

Flowchart for the longitudinal analysis CHARLS: the China Health and Retirement Longitudinal Study.

Blood sample assessment

In wave 2011 and 2015, fasting blood samples were tested. These blood samples were transported to Peking University and preserved at – 20 °C until transported to the Chinese Center for Disease Control and Prevention (CCDCP) within 2 weeks for testing. The blood testing mainly contained biochemical test and blood routine examination. Indicators in the present study involved high-density lipoprotein cholesterol (HDL-C), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), creatinine, blood urea nitrogen (BUN) and C-reactive protein (CRP), hemoglobin (Hb), white blood cell (WBC) count and platelet count, etc.

Definition of RC and hyperuricemia

Lipid related indicators such as high-density lipoprotein cholesterol (HDL-C), total cholesterol (TC), LDL-C were extracted. RC was calculated through the following formula: RC (mg/dl) = TC (mg/dl) – HDL-C (mg/dl) – LDL-C (mg/dl)²³. According to the Consensus of multidisciplinary experts on diagnosis and treatment of hyperuricemia related diseases in China²⁴, hyperuricemia was defined as the serum uric acid (SUA) ≥ 357 μmol/L (6 mg/dL) for females and ≥ 417 μmol/L (7 mg/dL) for males. Compared with another diagnostic criteria (SUA ≥ 360 μmol/L for females or SUA ≥ 420 μmol/L for males), this is a more precise definition of hyperuricemia in China^25–27.

Covariates

Covariates in this study consist of three parts: demographic characteristics, laboratory findings and background comorbidities. The demographic information and classification were listed as bellow: age (≥ 60 or < 60 years old), sex (male, female), marital status (married or unmarried), body mass index (BMI, ≥ 28 or < 28 kg/m²), and lifestyle such as sleep time (< 6 h, 6–8 h, > 8 h), alcohol or cigarette consumption (yes or no); Laboratory findings classification were listed as bellow: creatinine (normal or high: bounded by 1.35 mg/dL in male or 1.04 mg/dL in female), BUN (≥ 20 mg/dL or < 20 mg/dL), CRP (bounded by the median value), Hb (normal or low: classified by 120 g/L in male or 110 g/L in female), WBC (≥ 10 × 10⁹/L or < 10 × 10⁹/L) and platelet (≥ 100 × 10⁹/L or < 100 × 10⁹/L). Covariates of chronic comorbidities were all divided into two groups (yes or no), mainly involved the common chronic diseases in Chinese like hypertension, stroke, chronic lung diseases, hepatic disease, renal diseases and digestive diseases, Hepatic disease included hepatitis, liver cyst, hepatic aneurysm, etc. except for tumors and cancer. Renal disease included renal calculus and CKDs etc., but excluding tumor or cancer. Digestive diseases covered digestive ulcer, gastritis, etc. but not contained tumor or cancer. In addition, medicine treatment for hyperlipidemia (HLP) or hyperglycemia (HGC) was also adjusted.

Statistical analysis

In this study, continuous variables are expressed as median (interquartile range); categorical variables are exhibited as counts (percentage). Mann–whitney U test and Chi-square test were used to assess the baseline differences between the two groups.

Four logistic regression models were established to evaluate the association between RC and hyperuricemia. Model 1 was crude model; Model 2 was further adjusted for age, sex, marital status, BMI, sleep time and alcohol or cigarette consumption; Model 3 was further adjusted for creatinine, BUN, CRP, Hb, WBC and platelet; Model 4 was the full adjusted model which was further adjusted for hypertension, stroke, chronic lung diseases, hepatic disease, renal diseases, heart disease, digestive diseases and treatment for hyperglycemia and hyperlipidemia. Besides, tendency analysis and multiplicative interactional effect analysis were performed. Further, sensitive analysis was conducted. Sensitive analysis 1 were conducted in the participants without missing values in all covariates; Sensitive analysis 2 was done in the participants without overweight (BMI < 24 kg/m²); Sensitive analysis 3 was conducted in the participants without using blood hypoglycemic agent. In addition, stratification analysis of gender and age grade were conducted to ensure the robustness of the study.

1:1 PSM based on model 4 was adopted to reduce the confounders’ interference, in which the nearest-neighbor method was used, and the caliper was set at 0.03. Then, the propensity score and the standardized mean difference (SMD) was calculated. Sufficient balance between the two groups²⁸ was defined as SMD less than 0.01. Univariable and full adjusted multivariable regressions were then conducted. Additionally, IPTW (based on the full adjusted model) was applied to fully adjust for the potentially substantial difference in characteristics between groups. In this part, propensity score was developed, with the total covariates described above. Finally, propensity scores were used as the probability of a participant being assigned to exposed group and non-exposed group. IPTW was applied by weighting individuals using the inverse of their propensity scores to obtain a weighted population. After that, weighted logistic regression analysis were conducted.

RCS analysis is a method commonly used in medical studies for speculating nonlinear relationships between independent and dependent variables. It fits the curve by using cubic polynomials at different data intervals and smoothly connecting these polynomials at knots to fit the entire range of data. To explore whether the non-linear relationship exist between continuous RC and hyperuricemia, regression modeling strategies (rms) package was adopted to conduct RCS analysis (with 4 knots), and the regression was based on the full adjusted model.

In addition, CLPM analysis was adopted to identify causal relationship between RC and hyperuricemia in the middle aged and elderly Chinese population. CLPM was a longitudinal analytical model that investigated the inter-predictive relationship or quasi-causal relationship between variables²⁹. It reflects the dynamic effects between variables through cross-lagged paths, which constructs a path of autoregressive effect from the previous level of a variable to its current level, and a path of cross-lagged effect to the current level of another variable³⁰. This means the analysis of time-lag effect in CLPM allows to infer causality by analyzing variables’ changing in chronological order. Several standardized correlation coefficients were estimated to assess the fit goodness²⁹: comparative fit index (CFI), Tucker-Lewis index (TLI), root mean square error of approximation (RMSEA) and standardized root mean square residual error (SRMR). In these parameters, CFI > 0.9, TLI > 0.9, RMSEA < 0.08, SRMR < 0.08 indicated a good model fit. P < 0.01 was considered statistically significant.

Mediation analysis is used to determine whether the intermediary variable exist, and has a bridging role in the influence of dependent variable on independent variable. It further explains the mechanism of the influence of the dependent variable by differentiating the total effect of a treatment into direct and indirect effects. Indicators including total effect, direct effect, indirect effect and the proportion of mediating effect are produced in this process to evaluate the strength of mediation effect. In this paper, indirect associations mediated by WBC, CRP, hypertension and hyper glycemia were assessed. In this part, the “mediation” package in R was employed. RC and SUA were treated as continuous variables, and the analysis was based on the full adjusted multivariable logistic regression model.

Statistical analyses were carried out using Stata 16.1 (Stata Corporation, College Station, TX, United States) and the R 4.3.1 software (R Foundation for Statistical Computing, Vienna, Austria), with a significance level set at p < 0.05 (two-sided).

Results

Baseline characteristics of the study population

6,447 participants were enrolled in this study. 2958 participants were male, and 3489 participants were female. The baseline RC were 11.21 mg/dl in Q1 and 29.38 mg/dl in Q2 group, and the median of RC was 18.94 mg/dl. The prevalence of hyperuricemia was 6.4% in Q1 group and 11.9% in Q2 group. Participant in Q2 groups were more likely to have higher BMI. SUA of Q2 group tends to be higher in both survey 2011 and 2015. Platelet, WBC, CRP and HP also showed the similar results. Comorbidities such as digestive diseases, renal diseases, hepatic diseases and stroke has no difference between the two groups. While heart diseases exhibited statistical significance. The detail of baseline characteristics is displayed in Table 1.

Table 1.

Baseline characteristic files for the participants.

Covariates	Overall	Q1 (RC ≥ 18.94 mg/dl)	Q2 (RC < 18.94 mg/dl)	p
	6447	3224	3223
Age, years	58.00 [52.00, 65.00]	59.00 [52.75, 65.00]	58.00 [52.00, 64.00]	0.025
Gender (%)				< 0.001
Male	2958 (45.9)	1560 (48.4)	1398 (43.4)
Female	3489 (54.1)	1664 (51.6)	1825 (56.6)
Marital status (%)				0.319
With spouse	5459 (84.7)	2715 (84.2)	2744 (85.1)
Without spouse	988 (15.3)	509 (15.8)	479 (14.9)
BMI, Kg/m2	23.13 [21.22, 25.44]	22.79 [20.78, 24.46]	23.57 [21.71, 26.22]	< 0.001
Sleep time				0.691
< 6 h	3378 (52.4)	1687 (52.3)	1691 (52.5)
6–8 h	2585 (40.1)	1286 (39.9)	1299 (40.3)
> 8 h	484 (7.5)	251 (7.8)	233 (7.2)
Smoking (%)				0.753
No	4456 (69.1)	2222 (68.9)	2234 (69.3)
Yes	1991 (30.9)	1002 (31.1)	989 (30.7)
Drinking (%)				0.015
No	4346 (67.4)	2127 (66.0)	2219 (68.8)
Yes	2101 (32.6)	1097 (34.0)	1004 (31.2)
SUA (2015), μmol/L	279.79 [232.17, 333.37]	273.84 [226.21, 327.42]	291.70 [238.12, 339.32]	< 0.001
RC (2015), mg/dl	25.48 [19.69, 34.36]	22.78 [18.15, 29.34]	29.34 [22.01, 40.15]	< 0.001
SUA (2011), μmol/L	248.43 [208.32, 293.83]	242.98 [203.82, 287.48]	252.93 [212.52, 299.53]	< 0.001
RC (2011), mg/dl	18.94 [11.21, 29.38]	11.21 [7.35, 15.08]	29.38 [23.58, 38.66]	< 0.001
Platelet, × 109/L	207.00 [163.00, 258.00]	203.00 [161.00, 253.00]	211.00 [165.00, 262.00]	< 0.001
BUN, mg/dl	15.01 [12.49, 17.98]	15.24 [12.66, 18.37]	14.79 [12.32, 17.55]	< 0.001
Creatinine, mg/dl	0.75 [0.64, 0.86]	0.75 [0.64, 0.86]	0.75 [0.64, 0.87]	0.092
WBC, × 109/L	5.90 [4.90, 7.00]	5.70 [4.70, 6.71]	6.10 [5.17, 7.20]	< 0.001
CRP, mg/L	0.95 [0.53, 1.57]	0.85 [0.50, 1.43]	1.07 [0.58, 1.70]	< 0.001
Hyperuricemia				< 0.001
No	5858 (90.9)	3018 (93.6)	2840 (88.1)
Yes	589 (9.1)	206 (6.4)	383 (11.9)
Hepatic diseases (%)				0.071
No	6198 (96.1)	3085 (95.7)	3113 (96.6)
Yes	249 (3.9)	139 (4.3)	110 (3.4)
Renal diseases (%)				0.447
No	6035 (93.6)	3010 (93.4)	3025 (93.9)
Yes	412 (6.4)	214 (6.6)	198 (6.1)
Digestive diseases (%)				0.398
No	4935 (76.5)	2453 (76.1)	2482 (77.0)
Yes	1512 (23.5)	771 (23.9)	741 (23.0)
Lung diseases (%)				0.302
No	5787 (89.8)	2907 (90.2)	2880 (89.4)
Yes	660 (10.2)	317 ( 9.8)	343 (10.6)
Heart diseases (%)				0.001
No	5681 (88.1)	2885 (89.5)	2796 (86.8)
Yes	766 (11.9)	339 (10.5)	427 (13.2)
Treatment for HLP (%)				< 0.001
No	6098 (94.6)	3089 (95.8)	3009 (93.4)
Yes	349 (5.4)	135 (4.2)	214 (6.6)
Stroke (%)				0.654
No	6321 (98.0)	3158 (98.0)	3163 (98.1)
Yes	126 (2.0)	66 (2.0)	60 (1.9)
HP (%)				< 0.001
No	4893 (75.9)	2545 (78.9)	2348 (72.9)
Yes	1554 (24.1)	679 (21.1)	875 (27.1)
Treatment for HGC (%)				0.015
No	6207 (96.3)	3123 (96.9)	3084 (95.7)
Yes	240 (3.7)	101 (3.1)	139 (4.3)

All continuous variables were presented as median [IQR].

SUA serum uric acid, RC remnant cholesterol, BUN blood urea nitrogen, WBC white blood cells, CRP c-reactive protein, HLP hyperlipidemia, HGC hyperglycemia.

The longitudinal association between RC and hyperuricemia

In this part, 4 logistic regression models were established to explore the association between RC and the risk of hyperuricemia (Fig. 2). The ORs of hyperuricemia in the four models were 1.98 (1.66, 2.36), 1.92 (1.61, 2.31), 1.85 (1.55, 2.22) and 1.80 (1.50, 2.16) respectively. All p values and p values for trend were less than 0.001, which indicating the risk of hyperuricemia elevated with RC levels.

Fig. 2.

Forest plot illustrating the correlation between RC and hyperuricemia in logistic regression models. OR odds ratio, CI confidence interval.

In observational studies, data biases and confounding variables are commonly existed due to various reasons, which could influence the robustness of study. PSM and IPTW can alleviate the interference of these biases and confounding variables by calculating propensity score and inverse probability weighting, so as to make a more reasonable comparison between the experimental group and the control group. In this study, the results of PSM and IPTW were shown in Fig. 3. 2922 pairs of participants were matched after PSM. Figure 3A displayed the matching status between the treated and raw groups, indicating that the distribution of propensity scores concentrated after PSM, and the two groups were well matched. All SMD were less than 0.1. In addition, IPTW was applied to adjust the balance between the two groups. The SMD before and after IPTW was shown in Fig. 3B. All SMD were less than 0.05 after IPTW. Figure 3 indicating the covariates were well balanced after data processing. The outcomes of logistic regression were shown in Fig. 4. After PSM, the OR was 1.88 (1.56, 2.27) in crude model, and in the full adjusted model (adjusted as Model 4) was 1.86 (1.54, 2.25). These results further suggested that the risk of hyperuricemia increased with high RC levels. The results exhibited the same trend in after IPTW.

Fig. 3.

The matching conditions of covariates after propensity score matching and inverse probability weighted processing. (A) Histogram elucidating the matching status of propensity score after score matching; (B) Scatter graph displaying the SMD after inverse probability weighted processing.

Fig. 4.

Forest plot displaying the association between the remnant cholesterol and hyperuricemia OR odds ratio, PSM propensity score matching, IPTW inverse probability of treatment weighting, CI confidence interval, crude model: univariate logistic regression; adjusted model: the covariates were full adjusted as model 4.

Taken together, it was strengthened that high level of RC in 2011 is positively associated with the risk of hyperuricemia in 2015.

Results of sensitive analysis

The results of sensitive analysis were shown in Table 2. After excluding the interference of missing values, sensitive analysis 1 still exhibited apparently increased ORs in high RC group in all the four models. Particularly, the ORs were 1.83 (1.44, 2.32), 1.88 (1.48, 2.38), 1.80 (1.42, 2.29) and 1.77 (1.39, 2.26) in sensitive analysis 2, which suggesting the conclusion still hold in the participants without overweight. Hypoglycemic agent may affect the serum uric acid. To reduce the influence of hypoglycemic agent, sensitive analysis 3 was conducted only in the participants without using blood hypoglycemic agent, and the result was the same as before. Results of stratification analysis were shown in Table 3. Generally, the hyperuricemia risk elevated with RC levels both in the old participants (aged over 60 years old) and the middle-aged participants. The same trend was observed in the male and female stratification.

Table 2.

Results of sensitive analysis.

Sensitive analysis	Model1	Model2	Model3	Model4
1	1.93 (1.47, 2.55)***	1.91 (1.45, 2.53)***	1.82 (1.38, 2.42)***	1.76 (1.33, 2.35)***
2	1.83 (1.44, 2.32)***	1.88 (1.48, 2.38)***	1.80 (1.42, 2.30)***	1.77 (1.39, 2.26)***
3	1.99 (1.66, 2.38)***	2.04 (1.70, 2.45)***	1.94 (1.62, 2.34)***	1.87 (1.56, 2.26)***

Sensitive analysis 1: conducted in the participants without missing values in all covariates; Sensitive analysis 2: done in the participants without overweight (BMI < 24 kg/m2); Sensitive analysis 3: conducted in the participants without using blood hypoglycemic agent.

Table 3.

Stratification analysis of age and gender.

Covariates		Model 1	Model 2	Model 3	Model 4
Age (years)	≥ 60	1.77 (1.39, 2.28) ***	1.8 (1.4, 2.31) ***	1.72 (1.34, 2.23) ***	1.62 (1.26, 2.1) ***
	< 60	2.23 (1.74, 2.88) ***	2.26 (1.76, 2.92) ***	2.09 (1.62, 2.71) ***	2.05 (1.59, 2.67) ***
Gender	Male	1.88 (1.47, 2.41) ***	1.89 (1.48, 2.42) ***	1.82 (1.42, 2.33) ***	1.79 (1.4, 2.31) ***
	Female	2.14 (1.66, 2.78) ***	2.16 (1.67, 2.8) ***	2.02 (1.56, 2.63) ***	1.87 (1.44, 2.45) ***

Exploration of the nonlinear relationship between RC and hyperuricemia

RCS model was employed to further explore the nonlinear association between RC and hyperuricemia. Figure 5 revealed that the risk of hyperuricemia apparently increased with the continuous RC (p for overall < 0.001), while no significant nonlinear relationship was found between RC and hyperuricemia (p for non-linearity = 0.125).

Fig. 5.

Non-linear relationship between remnant cholesterol and hyperuricemia reflected by restricted cubic spline OR odds ratio, CI confidence interval.

Results of CLPM

CLPM was used to explore the possible causal relationship between RC and BUA. Figure 6 showed the cross-lagged model built on continuous RC and BUA at time points 2011 and 2015. All coefficients were standardized. RC at wave 2011 were found to induce higher BUA at wave 2015 (β = 0.005, P < 0.001). While the effect of BUA (wave 2011) on RC (wave 2015) had no statistical significance. In this model, CFI and TLI were approximately equal to 1.00, and RMSEA and SRMR nearly equaled to zero, which indicating that the model fitted well. In line with the outcomes of the longitudinal logistic regression analysis, this result suggested high level of RC could cause BUA increasing correspondingly.

Fig. 6.

Coefficients of cross lagged panel model between remnant cholesterol and blood uric acid. All coefficients were standardized; RC remnant cholesterol, BUA blood uric acid.

The results of mediation effect analysis

The mediation effect of WBC, CRP and hypertension were detected in wave 2011. Results showed that all the three mediators had indirect effect on the positive association between RC and BUA (Fig. 7). The mediated proportion of WBC was 3.40% (p = 0.010). CRP were found has a stronger mediating effect, which account for the 6.60% (p < 0.001). Hypertension showed a weaker mediating effect but still has statistically significant (3.23%, p = 0.004). This indicating that RC could not only direct cause high levels of BUA, but affecting the risk factors to induce hyperuricemia.

Fig. 7.

The mediating effect of white blood cell, C-reactive protein and hypertension to the relationship between remnant cholesterol and blood uric acid. (A) Coefficients of mediating effect of white blood cell (B) Coefficients of mediating effect of C-reactive protein (C) Coefficients of mediating effect of hyperuricemia; All the analysis was based on model 4. The mediating effect was labeled in red.

Discussion

This study discovered a strongly positive relationship between RC levels and hyperuricemia in Chinese aged over 45. Besides, a causal relationship was identified between RC and uric acid. WBC, CRP and hypertension participate in mediating this relationship. These findings indicating RC may play a role in the pathological processes of hyperuricemia, and may be a new point to prevent or treat hyperuricemia. To our knowledge, this is the first time to report the longitudinal association between RC and hyperuricemia.

RC predominantly comprise very low-density lipoproteins, chylomicron remnants, and intermediate-density lipoprotein³¹. As an atherogenic lipoprotein, it is mainly studied in the diseases related to lipid metabolism disorders. In recent years, evidences have been reported on the promoting effect of RC to atherosclerosis^32,33 and cardiometabolic diseases^34,35. Besides, RC also participates in the pathological process of stroke, obesity, frailty and even rheumatoid arthritis, etc.^36–39. But it is rarely studied in the hyperuricemia. The existing research is only a few cross-sectional studies, and the causal relationship was not testified⁴⁰.

In this paper, we demonstrated that the risk of hyperuricemia in high RC groups increased to 1.8 times after full adjusted, and even after PSM, the risk was still increased to 1.86 times compared with the groups with low RC. Besides, the OR was still 1.80 in the weighted logistic regression, which suggesting RC was an independent risk factor for the hyperuricemia incidence. These findings were partially in accordance with the previous studies of Xiaohai Zhou⁴¹. Moreover, the causal relationship between the RC and BUA was confirmed in our study. One possible explanation to elucidate this positive association was that RC was positively correlated with BMI, waist circumference, triglyceride, etc. and negatively correlated with HDL-C, which had been proved as the stimulating factors to hyperuricemia^42–44. On the other hand, very low-density lipoprotein (VLDL) was one of the main component of RC, and VLDL has been proved commonly elevated in patients with gout and asymptomatic hyperuricemia^45,46; and VLDL-C was an independent risk factor to hyperuricemia⁴⁷, which indicating a direct effect of RC to hyperuricemia.. Moreover, the synthesis of VLDL from triglyceride involves a variety of biological process, in which the fatty acid was produced and accumulated due to the triglyceride hydrolyzed by adipose triglyceride lipase⁴⁸; this could accelerate the decomposition of adenosine triphosphate, and eventually increased the production of uric acid⁴⁹.

In this paper, sensitive analysis was conducted to find the differences under different conditions. With ruling out the interference of missing data, the positive association still hold. The reliability and stability of the conclusion was further proved in this step. It is well known that obesity is often accompanied by abnormal lipid metabolism^50,51, and this may affect the RC level. Therefore, it is important to explore whether RC is positively related to hyperuricemia after excluding the interference of obesity. Our results showed that even in the participants without overweight, elevated RC level still linked to the higher risk of hyperuricemia. This may suggest that RC should be intervened even in the population without overweight. Previous study has documented the blood hypoglycemic agent such as insulin, acarbose, glibenclamide can stimulate the synthesis and secretion of BUA^52–54. So, we testified the association between RC and BUA in the participants without using hypoglycemic agent, and the results exhibited the similar elevated risk of hyperuricemia in group with increased RC.

As previous described, inflammation interacted with the abnormal lipid metabolism^55–57. Thus, the mediating effect of CRP and WBC were explored. And we demonstrated that CRP and WBC indeed mediated the positive association between the RC and uric acid concentration. Meanwhile, RC was positively correlated with RC and WBC, which was in accordance with the Yuxuan Wu’ research⁵⁸. Hypertension could enhance the development of hyperuricemia⁵⁹. Moreover, it has been certified that high RC level potentiates the development of hypertension. So, the link among RC, hypertension and hyperuricemia was tested in this paper by mediation analysis, and the results showed hypertension also has the mediating effect to the correlation between RC and hyperuricemia. All these put forward the related possible mechanisms of RC facilitating the occurrence of hyperuricemia.

In conclusion, this article revealed RC is positively associated with the prevalence of hyperuricemia in the middle aged and elderly Chinese, and proved high level of RC is an independent risk factor for hyperuricemia. Importantly, it is still fit in the participant without overweight. The underlying mechanism may be related to the inflammation and hypertension. RC is an indictor easily to acquire with low cost, as the components were the routine items in blood biochemistry examination. It enriched the predicting methods for hyperuricemia, and provide the basis for further basic experimental studies.

In this research, longitudinal analysis was conducted to investigate the association between RC and hyperuricemia, limitations still should be noted. First, some covariates were based on self-reports. Despite this method has been reported practicably in previous research⁶⁰, disadvantages still existed. For example, participants may give the fuzzy information because of lacking in medical knowledge; this could cause missing values and interfere the statistical analysis. Second, this is a study from Chinese middle-aged and elderly people, which might affect the generalizability of the findings. Thus, the conclusion needs to be validated in different populations from other countries; In addition, it would be more convincing to conduct the cross-lagged mediation analysis, which needs blood test at least for three waves; while till now, only two waves (2011 and 2015) contain blood sample test in CHARLS. The third, this is a retrospective observational study, so prospective clinical study should be designed to cover the deficiency of retrospective study. At last, the related molecular biological mechanism remains unclear, animal experiments are need for further exploration, which may be a new promising research point.

Despites of these shortages, this research still support elevated RC could lead to hyperuricemia, and the process is mediated by inflammation and hypertension. It reminds that we should pay more attention to serum RC level not just lipid level to avoid hyperuricemia.

Author contributions

Yanyuan Zhang analyzed the data and wrote the main manuscript text, Feifei Xu and Jin Ma prepared Figs. 1, 2, 3, 4, 5, 6, 7 and Tables 1, 2, 3. Yanyuan Zhang, Jin Ma and Feifei Xu participated in the study design. All authors reviewed the manuscript.

Data availability

Sequence data that support the findings of this study are available in the CHARLS repository (http://charls.pku.edu.cn/).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participation

The CHARLS study was approved by research ethics committees of Peking University. All participants provided written informed consent. No experimental interventions were performed. All procedures in this study were in accordance with the Declaration of Helsinki (revised in 2013) and were approved by the Ethics Committee of Peking University (IRB 00001052–11014).