Associaton of Retinol Binding Protein 4 (RBP4) Levels With Hyperuricemia: A Cross-Sectional Study in a Chinese Population

Background There are few studies on predictive biomarkers for hyperuricemia, and the predictive value of these biomarkers tends to be poor. Additionally, no reports have described the predictive value of retinol binding protein 4 (RBP4) for hyperuricemia. Purpose This study was performed to evaluate the value of RBP4 for predicting the risk of hyperuricemia in a general population, determine whether RBP4 could be used alone or in combination with other factors to predict the risk of hyperuricemia in the general population, and establish an optimum predictive model. Methods We conducted a population-based cross-sectional survey in 2018, involving a questionnaire, physical examination, and laboratory testing. We enrolled 2303 individuals by stratified random sampling, and 2075 were included in the data analysis after applying the eligibility criteria. Results Serum RBP4 level had a highly significant association with hyperuricemia (P<0.001). After adjusting for potential confounders, logistic regression indicated that the risk of hyperuricemia was highest in the highest RBP4 quartile (odds ratio: 7.9, 95% confidence interval [CI]: 4.18–14.84, compared to the lowest quartile). The area under the receiver operating characteristic (ROC) curve (AUC) for RBP4 was 0.749 (95% CI: 0.725–0.774, P<0.001), which was higher than that for all the other predictors assessed. The optimum model for predicting hyperuricemia in the general population consisted of RBP4, sex (male), body mass index, serum creatinine, high-sensitivity C-reactive protein, fasting blood glucose, insulin, and alcohol consumption. The AUC was 0.804 (95% CI: 0.782–0.826, P<0.001). Conclusions RBP4 is strongly associated with hyperuricemia, and its predictive value was higher than that of traditional predictors.

Unfortunately, studies on the risk and predictors of hyperuricemia are rare. Although some groups have reported on factors (age, body mass index [BMI], waist circumference [WC], triglycerides [TG], etc.) (7,17) that influence hyperuricemia, few studies have focused on individual biomarkers (visceral adiposity index (18), TG-glucose index (19), etc.) or models for predicting hyperuricemia (20)(21)(22). Additionally, most reported predictors have poor performance or are not suitable for large-scale clinical use because of issues such as cost and complexity.
Retinol binding protein-4 (RBP4) was recently recognized as a type of adipokine (23). Several small-sample studies have reported an association between RBP4 and hyperuricemia in patients with diabetes, chronic kidney disease, and MetS, but none of them focused on the general adult population (24,25). To our knowledge, there are no studies on the predictive value of RBP4 for hyperuricemia.
This large-scale study aimed to explore the association between RBP4 and hyperuricemia in the general population, and to evaluate whether RBP4 can predict hyperuricemia. We also hope that in the absence of serum uric acid results, abnormally elevated levels of RBP4 may suggest the importance of further serum uric acid testing.

Study Population
A cross-sectional survey of the general population in Wanzai town in Zhuhai city, China, was conducted in 2018. Using stratified random sampling, 2303 adults who were permanent residents of Wanzai town were initially enrolled in the study [1054 in the first visit in 2018 and another 1249 later in 2018 (the second visit was the follow-up visit of a similar epidemiological survey that we conducted in 2012 in Wanzai town)]. We followed the same sampling procedures described in previous reports on our 2012 epidemiological survey (26)(27)(28)(29)(30).
The exclusion criteria were as follows: (1) treatment with uric acid-lowering drugs or drugs that affect uric acid in the last 6 months, (2) severe liver or kidney damage, (3) acute cardiovascular or cerebrovascular diseases, and (4) pregnancy or breastfeeding.
After applying the eligibility criteria, 2075 individuals were included in the final analysis.

Ethics Approval
The study was approved by the Ethics Committee of the Third Affiliated Hospital of Southern Medical University and was conducted in accordance with the Declaration of Helsinki. All subjects provided written informed consent at recruitment.

Data Collection
Data on sociodemographic factors (age, sex, and education level), lifestyle (physical inactivity [physical activity <1 time/week], current smoking, and current alcohol consumption [≥1 time/ week]), medical history (hypertension, diabetes, coronary heart disease [CHD], and stroke), and medication use were collected using a questionnaire. The detailed design of this study (including the urine specimen collection) was the same as in our previous epidemiological survey conducted in 2012 (26)(27)(28)(29)(30).
Physical examinations were performed to collect data on blood pressure (systolic blood pressure [SBP] and diastolic blood pressure [DBP]), weight and height (which were used to calculate BMI), and WC. BMI was used to define obesity according to Chinese obesity criteria. Normal weight was defined as BMI<24 kg/m 2 . Obesity was defined as BMI >=28 kg/m 2 , BMI 24-28 kg/m 2 was overweight.

Statistical Analysis
Continuous variables with a normal distribution are presented as the mean and standard deviation, while those with a non-normal distribution are presented as the median and interquartile range. Categorical variables are presented as frequencies and Abbreviations: ACR, albumin-to-creatinine ratio; b2MG, b2 microglobulin; BMI, body mass index; CHD, coronary heart disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FBG, fasting blood glucose; HDL, highdensity lipoprotein; HOMA-IR, homeostasis model assessment of insulin resistance; hs-CRP, high sensitivity C-reactive protein; IL-6, interleukin 6; LDL, low-density lipoprotein; NAG, N-acetyl-b-D-glucosaminidase; RBP4, retinol binding protein 4; SBP, systolic blood pressure; SCr, serum creatinine; TG, triglycerides; WC, waist circumference. percentages. Continuous variables were compared between groups using independent-samples t tests or analyses of variance (for normally distributed variables) or Mann-Whitney U tests (for non-normally distributed variables). Categorical variables were compared between groups using c 2 or Fisher's exact tests.
Six binary logistic regression models (stepwise conditional), with hyperuricemia as the independent variable and RBP4 quartile as the independent variable, were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). The regression models adjusted for the following covariates: (1) age and sex; (2) model 1 covariates plus hypertension, diabetes, CHD, education of high school or above, physical inactivity, current smoking, and current alcohol consumption; (3) model 2 covariates plus SBP, DBP, log TG, LDL, HDL, BMI, eGFR, FBG, and log insulin; (4) model 2 covariates plus SBP, DBP, LDL, HDL, BMI, eGFR, and log HOMA-IR; (5) model 4 covariates plus log hs-CRP, and log IL-6; and (6) model 5 covariates plus log NAG and log ACR. SCr is strongly correlated with eGFR, so it was not included in the regression models.
The area under the receiver operating characteristic (ROC) curve (AUC) was used to assess the predictive value of RBP4, other predictors, and the following three predictive models for hyperuricemia: (1) RBP4, sex, BMI, SCr, log hs-CRP, log insulin, log HOMA-IR, and current alcohol consumption; (2) RBP4, sex, BMI, SCr, log hs-CRP, and FBG; and (3) RBP4, sex, BMI, SCr, log hs-CRP, FBG, log insulin, and current alcohol consumption. Youden's index (sensitivity + specificity -1) was used to select the optimum cutoff value of RBP4. The reciprocals of HDL and eGFR were used due to their negative associations with serum uric acid level.
Statistical analyses were performed in SPSS software version 20.0 (IBM Corp., Armonk, NY, USA). A two-sided P value <0.05 was considered statistically significant.

RESULTS
The baseline characteristics of the participants are shown in Table 1. Those in the hyperuricemia group were more likely to be male and older compared to those in the non-hyperuricemia group, and they had higher rates of hypertension, diabetes, CHD, alcohol consumption, and physical inactivity (P<0.05). In addition, there were higher values of SBP, DBP, WC, BMI, TG, LDL, FBG, insulin, HOMA-IR, SCr, cystatin C, NAG, hs-CRP, IL-6, and RBP4 in the hyperuricemia group compared to the non-hyperuricemia group (P<0.05), but lower values of HDL, eGFR, and serum uric acid (P<0.001).
The participants were divided into quartiles based on RBP4. Those in the fourth quartile were more likely to be male and older compared to the participants in the first quartile, and they were less educated and had higher rates of hypertension, diabetes, smoking, alcohol consumption, and physical inactivity (P<0.05). The values of SBP, DBP, WC, BMI, TG, LDL, FBG, insulin, HOMA-IR, serum uric acid, SCr, cystatin C, NAG, hs-CRP, IL-6, and ACR were higher and the values of HDL and eGFR were lower in the fourth quartile compared to the first quartile ( Table 2).
The prevalence of hyperuricemia gradually increased from the first to the fourth RBP4 quartile from 5.0% to 58.2% (7.5% to 61.7% in males and 4.4% to 55% in females) (all P<0.001, Table 3).
The associations between RBP4 and hyperuricemia, according to six multivariate binary logistic regression analyses, are shown in Table 4. RBP4 and hyperuricemia were positively associated in all six models. In the final model (model 6), the OR comparing quartile 4 of RBP4 with quartile 1 was 7.90 (95% CI: 4.18-14.84; P < 0.001).

DISCUSSION
Using population-based data on southern Chinese adults collected in a single-center cross-sectional epidemiological survey, we found that RBP4 had a highly significant association with hyperuricemia (P<0.001). More importantly, RBP4 was a good predictor of hyperuricemia; indeed, it performed better than traditional predictors. We also explored the predictive value of RBP4 combined with routinely assessed clinical factors that are related to hyperuricemia. To our knowledge, this is the first study to investigate the predictive value of RBP4 for hyperuricemia.
Several groups have reported individual predictors or predictive models for hyperuricemia (20)(21)(22). Lee MF et al. (20) reported that sex, BMI, and peroxisome proliferatoractivated receptor (PPAR)-g polymorphism are good predictors of hyperuricemia. Cao et al. (21) developed sex-specific prediction models (the predictors for males were age, SBP, BMI, and blood uric acid, and the predictors for females were SBP, BMI, TG, and blood uric acid) for hyperuricemia. Lee S et al. (22) developed a machine learning model (involving basic healthcare checkup test results) for predicting hyperuricemia. In addition, various novel blood lipid indicators such as the visceral adiposity index (18,33), TG-glucose index (19), lipid accumulation products (34), and the cardiometabolic index (35) have been independently associated with hyperuricemia. However, all these studies were limited to specific populations, and the results were not verified in other regions such as the southern China. RBP4, as a single factor, had a good performance in predicting the risk of hyperuricemia. The AUC for RBP4 in the general population was 0.749. This is better than previous traditional individual predictors and similar to the predictive model developed by Lee MF et al. (20), which had an AUC of 0.775. We also obtained a predictive model (model 3), involving RBP4 and traditional predictors, with a good AUC of 0.804. Notably, the predictive value of RBP4 was slightly better in females than males (AUC for RBP4: 0.738 vs. 0.756; AUC for predictive model 3: 0.782 vs. 0.824).
The conclusion of our investigation is consistent with two previous studies (24,25). Chen et al. were the first to report elevated serum RBP4 levels with increasing serum uric acid among 885 individuals with type 2 diabetes in Taiwan. Thereafter, Chan et al. (25) reported that serum RBP4 was positively associated with uric acid in 26 subjects with hypertension and MetS.
Our study has the following strengths. First, the participants were randomly sampled from the general population and covered all ages of adults, so the sample is widely representative. Second, 2075 people were included in the analysis, far exceeding the sample sizes of previous studies. Data are shown as mean±standard deviation, median (interquartile range), or frequency (percentage). ACR, urinary albumin-to-creatinine ratio; b2MG, b2 microglobulin; BMI, body mass index; CHD, coronary heart disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FBG, fasting blood glucose; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessment of insulin resistance; hs-CRP, high-sensitivity C-reactive protein; IL-6, interleukin 6; LDL, low-density lipoprotein; NAG, N-acetyl-b-D-glucosaminidase; RBP4, retinol binding protein 4; SBP, systolic blood pressure; TG, triglycerides; WC, waist circumference.
Third, strict eligibility criteria were adopted. For example, we excluded individuals who were treated with uric acid-lowering drugs and those with a history of diseases or medications that might affect the RBP4 level. Fourth, the association remained significant after adjustment. For example, after adjusting for 21 potential confounders, the final regression model still showed that the risk of hyperuricemia in the highest RBP4 quartile was still 7.9 times higher than that in the lowest quartile, indicating a significant independent association between RBP4 and hyperuricemia. Finally, this is the first study to evaluate the value of RBP4 for predicting the risk of hyperuricemia, and we found that RBP4 alone performed better than traditional

0.23
Data are shown as mean±standard deviation, median (interquartile range), or frequency (percentage). ★ vs. quartile 1, P < 0.05; ▲ vs. quartile 2, P < 0.05; ◆ vs. quartile 3, P < 0.05. predictors. This is also the most important finding of this study, as it indicates that RBP4 might be useful for predicting the risk of hyperuricemia and could be used in epidemiological research on hyperuricemia in general adult populations. Though direct assessment of uric acid should be prefered over measuring a surrogate, the association between RBP4 and uric acid could be diagnostically exploited in case uric acid values are not available. The mechanisms by which how RBP4 predicts hyperuricemia remain unclear. RBP4 is considered independently related to insulin resistance, which is implicated in the pathogenesis of hyperuricemia (36,37). A recent study showed that RBP4 might be involved in hyperuricemia-induced insulin resistance by inhibiting IRS/PI3K/Akt phosphorylation (38). Even after adjusting for insulin level plus FBG, or HOMA-IR, RBP4 remained strongly associated with hyperuricemia. This indicates that RBP4 may be related to hyperuricemia through other mechanisms besides insulin resistance. This study also indicated that the mechanisms may not involve renal function.   When we adjusted for all renal function indicators (including glomerular and renal tubular function; i.e., eGFR, ACR, and NAG) in regression model 6, the regression results showed that the independent association between RBP4 and hyperuricemia remained significant. We speculate that besides insulin resistance, RBP4 may have additional unique non-renal function mechanisms such as pro-inflammatory effects and direct effects on vascular smooth muscle and uric acid (23,39,40). Further basic research on the mechanisms is needed in the future. Our results should be considered in the context of several limitations. First, the participants were all Han Chinese adults from Zhuhai city, and the results may not be generalizable to other ethnicities. In addition, this was a single-center study and therefore inevitably limited regarding the sample size; large multicenter studies are needed to verify the conclusions. Furthermore, the study was cross-sectional, and the underlying mechanisms were not explored in depth. Cohort or case-control studies and basic research on the underlying mechanisms should be performed to verify our findings.

CONCLUSIONS
This study found that RBP4 is significantly positively associated with hyperuricemia in adults and has good predictive value for the condition. Clinically, it can be used alone or in combination with other traditional indicators to predict the risk of hyperuricemia.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of the Third Affiliated Hospital of Southern Medical University. The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
G-BH, HZ, and C-XX participated in the research design, performance of the research, data analysis, and manuscript writing. X-FS, QZ, and X-SK participated in the research design and data analysis. P-CG, X-LL, JN, H-SC, and HX participated in the performance of the research and data analysis. All authors were involved in revising the manuscript and approved the final version.