Association of Serum Uric Acid with Lung Function in the National Health and Nutrition Examination Survey (NHANES), 2007-2012: A Cross-Sectional Study.

Background Evidence regarding the relationship between serum uric acid and lung function was controversial. Therefore, this study is designed to investigate whether serum uric acid was independently related to lung function in the National Health and Nutrition Examination Survey (NHANES) (2007-2012) after adjusting for other covariates. The present study was a cross-sectional study. The total participants from NHANES (2007-2012) were 30442. After exclusion of subjects, 9474 subjects remained for the nal analysis. The target independent variable and the dependent variable were serum uric acid measured at baseline and lung function respectively. Covariates involved in this study included age, sex, race, income-poverty ratio, body mass index, systolic blood pressure, diastolic blood pressure, blood urea nitrogen, cholesterol, creatinine, total protein, FeNO, calcium, alcohol drinking, smoke, phosphorus and total bilirubin. Besides, there is a threshold effect on the independent association between serum uric acid and FEV1 and PEF. Those results are only found in the general population. Further epidemiologic studies will still be required to conrm this reverse association between serum uric acid and lung function.


Background
Serum uric acid (sUA) is the nal breakdown product of purine degradation and present in the epithelial lining uid of the respiratory tract and in plasma [1,2]. Previous studies have suggested that elevated sUA were associated with cardiovascular diseases, including hypertension, stroke, coronary heart disease and congestive heart failure [3][4][5][6]. Similarly, there are also a variety of research to investigate the relationship between sUA and respiratory disease, including chronic obstructive pulmonary disease (COPD), pulmonary hypertension and obstructive sleep apnea and a signi cant correlation was found [7][8][9].
Lung function gradually decreases with the passage of time and varies greatly among individuals [10]. Individuals with an accelerated decline in lung function are more likely to suffer from chronic respiratory diseases and face a higher risk of all-cause death [11].
Therefore, it would be very necessary to identify biomarkers associated with this decline [12].
A few epidemiological studies have investigated the relationship between sUA and lung function in the general population [13][14][15][16][17]. However, the results have been controversial. In this study, we investigated the association between sUA and lung function in US population using data from the National Health and Nutrition Examination Survey (NHANES 2007(NHANES -2012.

Study Population
The National Health and Nutrition Examination Survey (NHANES) was a representative survey of the national population of the United States (US), which was designed and conducted by the National Center for Health Statistics (NCHS). The present study was a crosssectional study. The data analyzed were gained from the NHANES (2007)(2008)(2009)(2010)(2011)(2012). The NHANES methodological details are available at www.cdc.gov/nchs/nhanes/.
For participants aged <18 years, their parents/guardians furnished informed consent, and participants aged ≥18 years furnished informed consent on their own. The NCHS Ethics Review Board granted approval for the conduct of NHANES, and written informed consents were obtained from all participants.

Study Variables
The principal variables of this study were lung function (dependent variable, include FEV1, FVC and PEF) and sUA (independent variable). SUA were measured using a Beckman Synchron LX20 (Beckman Coulter, Inc., Brea, CA). Lung function was measured by Ohio 822/827 dry-rolling seal volume spirometers.
The following covariates were included: age, sex, race, income-poverty ratio, body mass index, systolic blood pressure, diastolic blood pressure, blood urea nitrogen, cholesterol, creatinine, total protein, FeNO, calcium, alcohol drinking, smoke, phosphorus and total bilirubin. Details of sUA and total lung function measurement process and other covariate acquisition process were available at www.cdc.gov/nchs/nhanes/.

Statistical Analyses
All estimates were calculated considering for NHANES sample weights. Our presentation of continuous variables was based primarily on whether they are normally distributed. If it was a normal distribution, we present the continuous variable as mean ± standard, and vice versa as the medium (Q1, Q3). Categorical variables were expressed in frequency or as a percentage. Weighted linear regression models (continuous variables) and weighted chi-square tests (categorical variables) were applied to calculate differences between different groups.
After adjustment for potential confounders, weighted multiple regression analyses were applied to calculate the independent relationship between sUA and lung function. Weighted generalized additive models and smooth curve ttings were applied to estimate the non-linearity of sUA and lung function. After adjusting for the same covariates in the linear regression models, two-piece wise linear regression models were further employed to examine the threshold effect of sUA on lung function. All the analyses were performed with the statistical software packages R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empowerstats.com, X&Y Solutions, Inc, Boston, MA). P values less than 0.05 (two-sided) were considered statistically signi cant.

Baseline characteristics of selected participants
We showed the description of weighted baseline characteristics of these nal selected population in table 1 according to quartiles of sUA. In general, the average age of the 9474 selected participants was 37.12 ± 16.03 years old, and about 49.19% of them were male.
Among different groups of sUA (quartiles, Q1-Q4), the variables presented in table 1 were all signi cantly different.
Association between sUA and lung function (FEV1, FVC and PEF) In this study, we constructed three models to analyze the independent effects of sUA on lung function (weighted multivariate linear regression). Crude Model, not adjusted; model I, age, sex and race were adjusted; model II, the covariates presented in Table 1 were all adjusted. The effect sizes (Odds ratio (OR)), 95% con dence intervals and P value were listed in Table 2. In the fully adjusted model, we For the purpose of sensitivity analysis, we converted the sUA from continuous variable to categorical variable (quartiles of sUA). The effect sizes of each quartile of sUA were all negative association with FEV, FVC and PEF in the fully adjusted model, which was consistent with the result when sUA was a continuous variable. The P for trend of sUA with categorical variables in fully adjusted model were also consistent with the result when sUA was a continuous variable. Besides, we also found the trend of the effect size in different sUA groups was non-equidistant (see table 2).
The results of nonlinearity of sUA and lung function In the present study, we analyzed the non-linear relationship between sUA and lung function ( Figure.2). Smooth curve and the result of Generalized additive model showed that the relationship between sUA and lung function was non-linear after adjusting for the covariates presented in Table 1. We used both linear regression and two-piecewise linear regression to t the association and select the best t model based on P for log likelihood ratio test.
The P for log likelihood ratio test of FEV1 and PEF was less than 0.05, which indicated a non-linear relationship between sUA and FEV1 and PEF. Therefore, we choosed two-piecewise linear regression for tting the association between sUA and FEV1 and PEF because it can accurately represent the relationship. By two-piecewise linear regression and recursive algorithm, we calculated the in ection point was 6.5mg/dl and 7.3 mg/dl for FEV1 and PEF respectively.  (see table 3 and Fig.2). We can see a sharp increase in the absolute value of the effect sizes after the lever of sUA was larger than the in ection point.

Discussion
In this study, we used a big and nationwide representative sample of US population. Our ndings indicated sUA was negatively associated with lung function (FEV1, FVC and PEF) after adjusting other covariates. Besides, we also found a non-linear relationship between sUA and FEV1 and PEF, and the effect sizes on the sides of the in ection point were not consistent. This result suggested a threshold effect on the independent association between sUA and lung function. When the level of sUA was larger than 6.5mg/dl or 7.3 mg/dl, the FEV1 or PEF sharply decreased than before.
Previous studies have estimated the association between sUA and lung function in the general population, however, the results have been inconsistent. Ahn KM et al. suggested that "increased sUA level was signi cantly associated with accelerated FEV1 and FVC" by a sample of 19237 participants [17]. Similar ndings were also reported in studies of Aida Y et al, Hong JW et al and Kobylecki CJ et al [13,15,16]. Their conclusions are consistent with our ndings. However, some other studies are inconsistent with our ndings. Song JU et al. reported that sUA may have a positive effect on lung function in middle aged healthy population [14]. We speculate that the reasons for the different results may be caused by the following factors: (1) the research population is different. Our study was a general population of US, while the study, which was inconsistent with our ndings, was targeted at middle aged healthy population of South Korea; (2) the different conclusion did not clarify the nonlinear relationship; (3) compared with our work, the study did not take into account the effect of FeNO, total bilirubin and income-poverty ratio on the sUA and lung function relationship when adjusting covariates. However, previous studies have con rmed that these variables were related to lung function [18-20].
The exact mechanism of this association between sUA and lung function is still unclear. The effect of uric acid on lung function appears to be a double-edged sword in vivo. Possible explanations for negative association between sUA and lung function are as follows. First, sUA have been shown to be risen in hypoxic states. Previous studies have suggested that sUA increase in hypoxic states, such as chronic heart failure and COPD [21]. It has also been pointed out that pulmonary hypoxia promotes purine catabolism, leading to increased production of sUA [22]. However, it is unclear whether the low-level hypoxia observed in the general population affects sUA levels, as our study excluded subjects with overt clinical disease, such as asthma, air ow limitation, or coronary heart disease. Second free radical inducing toxicity [26][27][28]. It is known that uric acid plays a major antioxidant protective role on the airway surface, thus protecting the airway from the effects of reactive species [29,30]. Uric acid from the human epithelial lining uid of the respiratory tract is thought to be co-secreted with mucus by submucosal nasal glands after uptake from plasma [1]. Subjects with su cient uric acid pools to combat oxidative stress may be protected from the decline in lung function caused by continuous exposure to oxidative stress.
These explanations still need to be treated with caution because the double-edged characteristics of sUA and the inconsistent results from previous studies make it di cult to conclude about whether sUA has a bene cial or noxious effect on lung function. Further research is still needed to investigate the molecular mechanism toward the relationship between sUA and lung function, given the importance of evidence to determine whether sUA concentration is an eligible diagnostic or prognostic indicator for related diseases.
The clinical value of this study is as follows. (1)  Our study has some strengths. (1) we addressed the nonlinearity in the present study and further calculated the in ection point, which could more accurately evaluate the relationship between sUA and lung function; (2) we adjusted more confounders, such as FeNO, total bilirubin and income-poverty ratio; (3) we handled target independent variable as both continuous variable and categorical variable and the results of two types of data were consistent. Such an approach can reduce the contingency in the data analysis and enhance the robustness of results. (4) our sample size is relatively large compared with previous similar studies.
Our work has several limitations. First, because this study has a cross-sectional design, it could not determine whether there was a causal relationship between sUA and lung function. Second, sUA level was measured at a single time point. Third, we did not take into account the use of uric acid-lowering drugs, but we exclude the population with gout. Finally, our research subjects are from NHANES

Conclusions
In conclusion, we nd that increased serum uric acid level was negatively associated with FEV1, FVC and PEF in a general population.
Besides, there is a threshold effect on the independent association between serum uric acid and lung function. When the level of serum uric acid was large than 6.5mg/dl or 7.3 mg/dl, the FEV1 or PEF sharply decreased than before. Those results are only found in the general population. Further epidemiologic studies will still be required to con rm this reverse association between serum uric acid and lung function.

Availability of data and materials
The datasets analyzed during the current study can be found in https://www.cdc.gov/nchs/nhanes/.   Model I, age, sex and race were adjusted.
Model II, age, sex, race, income-poverty ratio, BMI, systolic blood pressure, diastolic blood pressure, blood urea nitrogen, cholesterol, creatinine, total protein, FeNO, calcium, alcohol drinking, smoke, phosphorus and total bilirubin were adjusted.  Figure 1 Flowchart of the inclusion of participants.

Figure 2
Correlation between serum uric acid and lung function (FEV1, FVC and PEF). (a, c, e) Each black point represents a sample. (b, d, f) The area between two blue dotted lined is expressed as a 95% CI. Each point shows the magnitude of the serum uric acid and is connected to form a continuous red line. Age, sex, race, income-poverty ratio, BMI, systolic blood pressure, diastolic blood pressure, blood urea nitrogen, cholesterol, creatinine, total protein, FeNO, calcium, alcohol drinking, smoke, phosphorus and total bilirubin were adjusted.