Higher Atherogenic Index of Plasma Is Associated with Hyperuricemia: A National Longitudinal Study

Background The association between atherogenic index of plasma (AIP) and hyperuricemia remains indistinct. This study was aimed to examine the relationship between AIP and hyperuricemia among the middle-aged and the elderly Chinese population. Methods Datasets were retrieved from the China Health and Retirement Longitudinal Study (CHARLS) survey conducted in 2011 and 2015. 13,021 participants in the CHARLS in 2011 and 7,017 participants involved both in 2011 and 2015 were included, respectively. The measurement of AIP and hyperuricemia was based on the test of fasting blood. Association between AIP and hyperuricemia was assessed by logistic regression, and the nonlinear association was examined by restricted cubic splines (RCS). The cutoff point of AIP was calculated using receiver operator curve (ROC). 1 : 1 propensity score matching (PSM) was adopted to further explore the relationship between AIP and hyperuricemia. Results In the section of a cross-sectional study, a positive association between AIP and hyperuricemia was found. The odds ratios (ORs) of hyperuricemia were 1.00 (reference), 1.52 (1.10–2.10), 1.80 (1.31–2.47), and 3.81 (2.84–5.11). Nonlinear association was not detected using RCS analysis. There were 664 hyperuricemia cases during the four years follow-up. The hyperuricemia prevalence was 9.5%. In the fully adjusted longitudinal analysis, the ORs for hyperuricemia across the quartiles of AIP were 1.00 (reference), 1.00 (0.74–1.37), 1.59 (1.20–2.11), and 2.55 (1.94–3.35), respectively. In the longitudinal analysis after PSM, the OR of hyperuricemia were 1.91 (1.45, 2.51) and 1.92 (1.45, 2.54) in the univariate and multivariate model, respectively. Conclusion AIP can predict the prevalence of hyperuricemia in the Chinese middle-aged and elderly population.


Introduction
Hyperuricemia is a disease defned as an abnormal elevating status of plasma uric acid, which is the fnal enzymatic product of purine metabolism [1,2].Te main manifestation of hyperuricemia is gout, which can cause pathological pain in soft tissues or joints, and even result in renal dysfunction at the end stage [1].Nowadays, hyperuricemia has become a major public health problem that threatens human health, and its prevalence keeps rising constantly [3].It has been reported that the incidence rate of hyperuricemia in the US was up to 21.2% in male and 21.6% in female [4].Meanwhile, with the economic development in China, an increasing trend of hyperuricemia and gout has been observed.Literature documented that the overall hyperuricemia prevalence in China was about 13.3%, and the specifc morbidity ranges from 10.1% in northeast China to 18.6% in south China due to the demographic constitution and imbalanced economic development [5].Te rapid growth of hyperuricemia impaired the population health, and at the same time, it added the burden of the medical and healthcare system [6].Moreover, hyperuricemia was demonstrated closely correlated with some critical illness such as cardiovascular disease in recent years [7,8].A study from Korea reported hyperuricemia was associated with a nearly tripled risk for heart rate irregularity, which refecting total arrhythmia [9].Te risk and all-cause mortality of coronary heart disease rises when participants sufer from hyperuricemia [10].Besides, the high uric acid level can be applied to predict the severity of coronary artery disease [11].In addition, hyperuricemia also has relationship with a significant increased risk of hypertension and congestive heart failure [12][13][14].Tus, early prediction of hyperuricemia is worthy to be explored.
Besides, hyperuricemia has been confrmed to have an association with dyslipidemia, which is a cluster of blood-lipid disorder diseases.Similarly, dyslipidemia is also an independent risk factor for cardiovascular disease [15].Previous study testifed hypercholesterolemia is one of the principal driving forces in the development of atherosclerotic cardiovascular diseases [16].A large amount of evidence from genetic, epidemiologic, and clinical intervention studies has already frmed that high serum levels of low-density lipoprotein cholesterol cause the development of atherosclerotic plaque [17].Except that, hyperuricemia is often accompanied by abnormal lipid metabolism, including abnormal apolipoprotein A and low-density lipoprotein cholesterol levels; furthermore, the ratio of apolipoprotein B to A was found strongly associated with serum uric acid levels in US people [18].AIP, which is calculated as log (triglyceride/high-density lipoprotein), is an indicator for the process of lipid metabolism.It was proposed for the frst time by Dobiásová and has close correlation with lipoprotein particle size and esterifcation rate in apoB-lipoprotein-depleted plasma [19].Besides, it was also identifed as a novel indicator of cardiovascular disease among diferent populations [20].As an independent risk factor, AIP is useful for improving the diagnosis and prevention of subclinical coronary artery disease [21].A low level of AIP can even predict a decrease in allcause mortality of hospitalized patients with acute myocardial infarction [22], while the relationship between hyperuricemia and AIP has been rarely studied.
Given that uric acid has a close relationship with dyslipidemia, the association between the metabolism indicator AIP and hyperuricemia was examined in this article.To our knowledge, this was the frst research to identify the efect of AIP on the occurrence of hyperuricemia in Chinese middleaged and elderly people on a national scale.

Data Source and Analytical Sample.
In this paper, a dataset was retrieved from the CHARLS 2011 and 2015 surveys.CHARLS is a longitudinal project initiated from 2011 and followed up every 2 years in China.It is conducted by the Chinese Center for Disease Control and Prevention (CCDCP).Te project mainly involves home interviews, medical examination, etc., which were collected from 28 provinces and 150 counties in the whole of China.Information such as health information, demographics, socioeconomic status, as well as physical and physiological measurements was recorded.Specifc descriptions regarding the CHARLS could be accessed via the ofcial website (https://charls.pku.edu.cn/) or publications [23].
A total of 17,705 participants were enrolled in the 2011 baseline.After eliminating the missing data such as age, gender, or uric acid testing, 13,021 eligible participants were reserved for a cross-sectional study.Respondents who participated both in the CHARLS 2011 and 2015 waves were selected from CHARLS 2015.Besides, those with hyperuricemia at baseline in 2011 were excluded.At last, 7,017 participants were reserved for longitudinal analysis.Te detail is shown in Figure 1.

Data Collection and Covariates.
In the CHARLS survey, questionnaires containing basic information such as marital status, behavior, and history of diseases were handed out by experienced interviewers [23].Fasting venous blood from the respondents was acquired and centrifuged to gain plasma.Te plasma was then immediately stored under appropriate frozen conditions (−20 °C) until transporting to the CCDCP in Beijing for testing within 2 weeks.Te medical laboratory results were obtained by testing these specimens.Indicators including uric acid, triglyceride (TG), high-density lipoprotein (HDL), blood urea nitrogen (BUN), cystatin (Cyc) C, glucose, and low-density lipoprotein (LDL) were determined using enzymatic colorimetric tests [23].Te covariates in the present study consisted of age, gender, marital status, sleep duration, afternoon napping, cigarette consumption, alcohol consumption, body mass index (BMI), depression refected by the Center for Epidemiologic Studies Depression (CESD) scale-10 questionnaire, BUN, LDL, hepatic disease, renal disease, digestive disease, and arthritis disease.Age was categorized into 4 groups: 40-50 years, 50-60 years, 60-70 years, and >70 years; marital status was categorized into individuals living together or living alone.Sleeping time was classifed into three groups: sleeping less than 6 hours, between 6 and 8 hours, and more than 8 hours.Afternoon napping was defned according to whether participants had the custom of afternoon napping or not.Smoking status consisted of three statuses: currently smoking, never smoking, and quit smoking.Alcohol consumption was categorized as drinking more than once a month, less than once a month, and never drinking.BMI was divided into three groups: <18.5 kg/m 2 , <24 kg/m 2 , <28 kg/m 2 , and ≥28 kg/m 2 .Depression was assessed by CESD-10, and participants got score >10 were identifed as depressive state [24].LDL is classifed into two groups >120 mg/dl and <120 mg/dl.Hepatic disorders included hepatitis, liver cyst, and hepatic aneurysm except for fatty liver, tumors, and cancer.Renal disorders included kidney stones and CKDs excluding tumor or cancer.Digestive diseases covered digestive ulcer, gastritis, etc., but did not contain tumor or cancer.

Acquisition of AIP and the Defnition of Hyperuricemia.
TG and HDL of fasting blood plasma were examined as described above.Ten, AIP was calculated using the following formula [19]: AIP � log (TG(mg/dL)/HDL(mg/dL)); participants were classifed into four groups according AIP quartiles (Q): Q1 (−2.83, 0.29), Q2 (0.29, 0.77), Q3 (0.77, 1.30), and Q4 (1. 30, 5.33).According to the consensus of multidisciplinary experts on diagnosis and treatment of hyperuricemia-related diseases in China [25], the level of serum uric acid (SUA) to defne hyperuricemia was SUA Besides, subgroup analysis was employed to explore the association in particular groups.A restricted cubic spline was adopted to model and visualize the non-linear relationship between AIP and hyperuricemia index with 3 knots.To eliminate the possible infuence of potential confounders, a 1 : 1 propensity score matching (PSM) without replacement was performed based on the cutof point calculated from the ROC analysis.In the current investigation, 0.03 was selected as the caliper width for PSM.
In the PSM cohort, every possible confounder was considered [29].Longitudinal analysis was further used to access the association between AIP and hyperuricemia.A sensitive analysis was carried out using the full adjusted Model 5, only within the participants without use of lipid-lowering drugs.Figures were drawn by Stata and GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA).P < 0.05 (twosided) was considered statistically signifcant.

Demographic Characteristics of the Study Population.
A total of 13,021 participants at CHARLS 2011 baseline were included in the cross-sectional analyses.Among them, 6,146 were female and 5,367 were male.Across the quartiles, the mean (SD) of AIP is −0.07 ± 0.28, 0.54 ± 0.13, 1.02 ± 0.15, and 1.87 ± 0.54, respectively.Te mean age of the enrolled participants was 59.21 ± 9.71, and the age showed decreased trend across AIP quartiles.Table 1 shows that the number of individuals whose BMI <24 kg/m 2 apparently decreased across the quartiles of AIP, but the number increased across the quartiles in the participants with BMI ≥24 kg/m 2 .Te prevalence of currents moking was also lower among those with high AIP.It is less likely to have higher AIP in the participants without napping and with digestive disease.Te detail of baseline clinical profles attending the 2011 CHARLS survey is shown in Table 1.

Te Association between AIP Index and Hyperuricemia.
In the 2011 survey, the prevalence of hyperuricemia in the middle-aged and above Chinese was 5.20%.As shown in Table 2, the prevalence of hyperuricemia was 3.26%, 4.56%, 4.96%, and 10.55% in the four groups, respectively, which indicated an increasing trend across the quartiles of AIP.In Model Eligible participants for longitudinal analysis (n =7017) Figure 1: Flowchart of the study population.
International Journal of Endocrinology   3).Te baseline profles after PSM are shown in Supplementary Table 1 and Supplementary Figure 1.

Te Longitudinal Association between AIP and
Hyperuricemia.In this section, longitudinal analysis was performed in participants involved in both CHARLS 2011 and 2015 survey.Result showed that the overall prevalence of hyperuricemia in target population in 2015 was 9.5%.Across the quartiles of AIP, the incidence gradually increased and was 5.92%, 6.48%, 10.37%, and 15.48%.Besides, the increasing trend still exists in the ORs across the AIP quartiles in every model, and all p values for trend were less than 0.001 (Table 4).Signifcant subgroup efects on the association between AIP and hyperuricemia was not found in gender, renal disease, alcohol consumption, depression, and sleep duration.Among participants who never smoked, higher AIP was apparently associated with an increased risk of hyperuricemia; the OR was 1.78 (1.22-2.60)and 2.74 (1.90-3.95)compared with the frst quartile of AIP (Table 5).
PSM was used to reduce the interference of confounding factors.Figures 4(a) and 4(b) show that participants grouped by the cutof point of 0.96 well matched.Te cutof point was calculated from ROC, and the area under the curve was 0.67 (Figure 4(c)).After PSM, 2096 pairs of participants were reserved for further logistic regression.In longitudinal analysis, the ORs before PSM were 2. 28 4(d)).

Sensitive Analysis of Treatment for Dyslipidemia.
In this part, participants with use of lipid-lowering drugs were excluded.Tis sensitive analysis was carried out within Model 5. Similar with previous outcomes, the results still showed an apparently increasing risk of hyperuricemia with across the quantiles of AIP, which enhanced the robustness (Figure 5).

Discussion
Tis nationwide study estimated that approximately 5.2% of the middle-aged and above Chinese adults sufered from hyperuricemia in 2011, while the prevalence increased to 9.5% in the target population in 2015.AIP was demonstrated positively associated with the prevalence of hyperuricemia in the middle-aged and elderly Chinese.Tough there was no nonlinear relationship found, the curve of restricted cubic splines analysis generally exhibited a positive relationship between AIP and hyperuricemia, indicating AIP has a proper predictive ability for hyperuricemia.Tis positive association still existed after the adjustment for confounding factors by PSM.
Hyperuricemia has gradually become an important public health issue worldwide [30].Especially in China, an alarming rise was observed in recent years attributed to its growing economy and urbanization [6], which is in accordance with the fndings of this paper.It is worth noting that the prevalence of hyperuricemia varies in diferent geographic regions in China.Tis phenomenon might be related to variability in economic development, resulting in the lifestyle diference [5].A huge dietary transition occurred from relying on carbohydrates and vegetables to meat and dairy products, in which bacon, beef, lamb, and organ meats were purine-rich diet and identifed as the main pathogenic factors of hyperuricemia [31].Besides, alcohol was one dependent risk factor for hyperuricemia [32].Similarly, in our study, an apparent elevated risk for hyperuricemia was found in the participants drinking regularly than those drinking occasionally or never drink.Te risk of dyslipidemia increased due to the diet high in fat and protein [33].
Previous fndings also documented that hyperuricemia is positively associated with dyslipidemia in northwest China [28].Besides, a cross-sectional study in Chinese at 2020 revealed HDL-C is a protective predictor of serum uric acid levels in gout [34].Similarly, a signifcant negative correlation was observed between uric acid level and HDL-C among young adults in Qatar [35].In addition, a large retrospective cohort study revealed that TG levels independently afect the incidence of hyperuricemia [36].Te mechanism behind the increasing AIP and elevated TG levels is not fully clear.Te possible mechanism may be International Journal of Endocrinology  6 International Journal of Endocrinology mainly related to the following points.First, TG links with high uric acid may involve free fatty acids metabolic pathways, which is a physiological process for the long-term regulation of energy homeostasis [37].TG can take part in the synthesis of free fatty acids, in which the utilization of adenosine triphosphate in the process of re-esterifcation was accelerated.Elevate TG can lead to a high level of free fatty acid production and utilization, which could accelerate    International Journal of Endocrinology   International Journal of Endocrinology the decomposition of adenosine triphosphate, and this process increased the production of uric acid and causes its abnormal rise in blood [38].Second, many studies have confrmed that HDL-C is a protective factor against serum uric acid elevation in general [39][40][41].
Besides, a major protein component of HDL-C called apolipoprotein A-1 exerts an inhibitory efect on monocytes cytokines by reducing the activation of CD11b, since the monocytes can elevate serum uric acid levels through diferent pathways [42,43].Tus, it makes sense that AIP, in which TG serves as a numerator while HDL acts as a denominator, is positively correlated with hyperuricemia.
Even so, AIP was rarely studied in hyperuricemia.Duan et al. have proved hyperuricemia is an independent risk factor for high AIP levels a cross-sectional study at 2022 [44].But it made little sense to utilize a disease to predict an indicator.In the cross-sectional analysis part of our study, every single model demonstrated an apparently positive association between AIP and the prevalence of hyperuricemia.Especially in the fully adjusted model, the risk of hyperuricemia in the highest AIP quartile is 2.81 times bigger than the reference group.Further, RCS analysis also shows an upward trend of hyperuricemia with the AIP increasing.In addition, after attenuating the efect of confounding factors by PSM, both the univariate and multivariate logistic models were still robust; the risk of hyperuricemia increased 1.59 and 1.37 times, respectively.Besides, this trend still held in the sensitive analysis for the use of lipid-lowering drugs.A cross-Sectional study from northeast China suggested that AIP can predict hyperuricemia in rural population of northeast China [45], which was consistent with our fndings; besides, it showed that, compared to the low AIP group (<0.11), participants in increased AIP group (>0.21) had a 2.536-fold risk for hyperuricemia.In our study, after excluding the participants with hyperuricemia in 2011, the longitudinal analysis was conducted between CHARLS 2015 and 2011 survey.Similar outcomes showed that the risk of hyperuricemia apparently increased with the growth of AIP in all models compared with the reference group.Even after PSM, the outcome maintained accordantly.In addition, this longitudinal analysis research was based on a national scale survey of CHARLS, suggesting that AIP has the ability to predict the prevalence of hyperuricemia in the middle-aged and elderly Chinese population.
Generally, we found a practical indicator for predicting hyperuricemia prevalence, and it enriched the method alarming the hyperuricemia attack.In addition, AIP is a lowcost indicator because both TG and HDL are the basic items in the routine blood biochemical inspection on hospital admission or medical examination, which cost little to carry out.Besides, it is a feasible indicator to acquire, for it can be directly calculated and does not need more complicated physical examination and inspection.Given this, AIP is a high-quality and cost-efective indicator to alarm the occurrence of hyperuricemia, which has important clinical implications and may be a promising indicator in clinical practice.Nevertheless, further exploration is still demanded.To the best of our knowledge, this is the frst report to investigate the longitudinal relationship between AIP and hyperuricemia in the middle-aged and elderly Chinese population.
Our study also had several limitations.First, this is a retrospective study, which is limited to establish a causal relation between AIP and hyperuricemia.Terefore, largescale prospective studies are needed to verify the present fndings.Second, hyperuricemia is a disease interacted closely with diet, while the diet documented in CHARLS is not that sufcient, and this should be improved with the completeness of the database.Tird, the information collected from the participants may sufer from recall bias on self-report variables, so it still has chance to cause deviation.Finally, this study was based on the CHARLS survey in 2011 and 2015, which means it should be further verifed in other countries or regions worldwide.

Conclusion
Conclusively, this pilot study suggested that AIP can predict the prevalence of hyperuricemia in Chinese middle-aged and elderly population.

Figure 2 :
Figure 2: Results of RCS analysis, ROC, and PSM.(a) RCS results for the prevalent hyperuricemia.(b) RCS results for the prevalent hyperuricemia in female.(c) RCS results for the prevalent hyperuricemia in male.Te RCS regression model was adjusted as Model 5. RCS: restricted cubic spline.(d) Receiver operator curve.(e) Kernel density plot before PSM.(f ) Kernel density plot after PSM.

Figure 3 :
Figure 3: Forest plot of logistic regression before and after PSM.Te full adjusted model as Model 5 was used in multivariate logistic regression.

Figure 4 :
Figure 4: Longitudinal analysis after PSM.(a) Kernel density plot before PSM.(b) Kernel density plot after PSM rather than ROC curve.(c) Receiver operator curve.(d) Forest plot of logistic regression before and after PSM.

Figure 5 :
Figure 5: Te sensitive analysis of treatment for dyslipidemia AIP: atherogenic index of plasma; OR: odds ratio; PSM: propensity score matching; Q: quartiles.

Table 1 :
Baseline clinical profles of participants attending the 2011 CHARLS survey.
N: number; Q: quartile; CHARLS: China Health and Retirement Longitudinal Study.

Table 2 :
Odds ratios (95% CI) for hyperuricemia by quartiles of AIP.Te Subgroup Analysis and Interaction Efect Analysis.Subgroup analysis was employed to further explore the association between AIP and hyperuricemia.Te increasing trend still existed in the subgroups of participants without renal disease and sleeping less than 6 hours.Subgroup diference was not observed among gender, renal disease, cigarette consumption, depression, and sleep duration, while drinking more than once a month was detected to exacerbate the association between AIP and hyperuricemia (P for interaction � 0.02).Besides, in these subgroups, all ORs at Q3 and Q4 generally exhibit an increasing trend.Details are exhibited in Table3.3.4.Results of RCS Analysis, ROC, and Logistic Regression after PSM.Restricted cubic splines analysis was employed to fexibly model and visualize the association between AIP and hyperuricemia.As shown in Figures2(a)-2(c), though there was no nonlinear relationship detected, all curves showed elevated trend of ORs with AIP increasing.P for nonlinearity was 0.45, 0.26, and 0.81, respectively, in total, male and female groups.P values for overall were all less than 0.001.
Logistic regression to examine the association between AIP index and hyperuricemia.Q1 was set as the reference group.Model 1: univariate regression; Model 2: adjusted by age, gender, marital status, sleeping time, and afternoon napping based on model 1; Model 3: further adjusted by alcohol and cigarette consumption, BMI and depression; Model 4: further adjusted by blood uric acid and LDL; Model 4: further adjusted by hepatic disease, digestive disease, renal disease, and arthritis disease.*p<0.05;* * p < 0.01; * * * p < 0.001.ROC displayed that the area under the curve was 0.73, indicating a good predictive power; and the cutof point of AIP was 1.05 (Figure2(d)).Figures2(e) and 2(f ) exhibited the Kernel density plot before and after PSM; outcomes showed that distribution of patients' propensity scores more concentrated after PSM, representing the AIP <1.05 group and AIP ≥1.05 group were well matched.After (1.94, 2.68) and 2.10 (1.74, 2.54) in univariate and multivariate logistic regression before PSM.Te baseline profles after PSM are shown in Supplementary Table 2 and Supplementary Figure 2.After PSM, the ORs were 1.91 (1.45, 2.51) and 1.92 (1.45, 2.54), respectively (Figure

Table 3 :
Subgroup analysis of association between quartiles of AIP and hyperuricemia.
*Association between quartiles of AIP and hyperuricemia stratifed by gender, renal disease, alcohol consumption, cigarette consumption, depression, and sleep duration.Q1 was set as the reference group.Models were adjusted for the same covariates as model 5 in Table2.Stratifcation variables were not adjusted in the corresponding models.Q: quartile; * p < 0.05; * * p < 0.01; * * * p < 0.001.

Table 5 :
Subgroups analysis of association between quartiles of AIP and hyperuricemia in longitudinal analysis.Subgroup analysis of association between quartiles of AIP and hyperuricemia.Models were adjusted for the same covariates as model 5 in Table2.Q1 was set as the reference group.Stratifcation variables were not adjusted in the corresponding models.Q: quartile; * p < 0.05; * * p < 0.01; * * * p < 0.001.