Relationships between serum total bilirubin levels and hypercholesterolemia in Chinese men and women.

Objective Bilirubin, an antioxidant, is reported to relate with metabolic syndrome and cardiovascular diseases. The current study is to assess the relationship between total bilirubin (TB) and hypercholesterolemia in the Chinese population. This cross-sectional study involved 48971 males and 31327 females who were in good health. Physical examinations and laboratory tests including TB and total cholesterol (TC) were performed. Subjects were divided into the following groups according to age: subjects younger than 30 years, subjects aged between 30 to 60 years and subjects older than 60 years. Binary logistic regression models examining factors independently related to TB was performed for males and females in different age groups. covariates.

Relationships between serum total bilirubin levels and hypercholesterolemia in Chinese men and women. Xuemei

Methods
This cross-sectional study involved 48971 males and 31327 females who were in good health. Physical examinations and laboratory tests including TB and total cholesterol (TC) were performed. Subjects were divided into the following groups according to age: subjects younger than 30 years, subjects aged between 30 to 60 years and subjects older than 60 years. Binary logistic regression models examining factors independently related to TB was performed for males and females in different age groups.

Results
TB was negatively related to most parameters in males and females, such as body mass index (BMI), blood urea nitrogen (BUN), TC, triglycerides (TG), low-density lipoprotein (LDL) and glucose (G). The incidence of high TC was higher in females than in males and decreased with the increasing of TB in both genders. People older than 30 years demonstrated a similar trend. When there were no other covariates, TB appeared to be an important risk factor for high TC for men and women. However, this relationship disappeared when other covariates were added in the binary logistic regression models.
Similar results also existed in men younger than 60 years and women younger than 30 years.

Conclusions
Our ndings implied that serum TB concentration was not an independent risk predictor for hypercholesterolemia. TB was thought to be a covariate factor in the development of dyslipidemia.

Background
Hypercholesterolemia, a metabolic disorder characterized by the elevated level of serum total cholesterol, is a common form of hyperlipidemia. Relationship between hypercholesterolemia and cardiovascular disease mortality was a research hotspot in recent years [1,2]. Literatures showed that hypercholesterolemia was associated with increased cardiac oxidative stress [3,4]. In mammalian tissues, TB was probably the most abundant endogenous anti-oxidant, accounting for the majority of human serum anti-oxidant activity [5]. Compared to other anti-oxidants, TB exhibited strong superoxide and peroxy radical scavenger activity [6]. Increasing in vivo evidences have found that TB played an important role in preventing many diseases (such as cardiovascular disease [7], Gilbert syndrome [8]) and reducing cancer mortality [9].
Recent studies have been committed to the relationship between diseases and their risk factors such as TB and oxidative stress. However, information is lacking on the interaction between TB and dyslipidemia. Previous studies reported a negative relationship between TB and TC. However, these studies have some shortcomings, such as narrow population distribution [10,11] and age range [12], gender restrictions [13],no further statistical analysis and the exclusion of other factors [9]. Opposite opinion was also existed. A cross-sectional analysis of 594 Indigenous Australians reported a positive correlation between TB and TC [14]. However, their research population was too small to be demographically representative.
Other articles found no relationship exist between TB and TC in healthy people [15][16][17]. The current study was performed to access the association between TB and TC in Chinese population.

Participants
Participants were recruited from a community-based health survey carried out by our multidisciplinary team in General Hospital A liated to Tianjin Medical University. Our previous articles had reported the protocol of that survey [18][19][20][21][22]. Participants were required to complete a detailed health questionnaire. On the basis of their responses, they attended a screening examination including taking blood after an overnight fast (time ≥ 8h). To control the confounding factors, subjects met the following situations were excluded: subjects with a medical history of blood, liver, kidney, stomach, thyroid in ammation, infection, cancer or autoimmune diseases; subjects taking any medicine which might affect TB and TC; subjects who had abnormal liver function tests, de ned as a alanine aminotransferase (ALT) level greater than 100 U/L or a serum TB level > 40 μmol/L; subjects who had excessive drinking or pregnancy. Visual hemolysis tests were performed on the centrifuged samples to avoid increased TB due to hemolysis.
Finally, a total of 48971 males and 31327 females were collected from September 2010 to September 2015.

Measures
Anthropometric measurements and fasting blood glucose tests were performed for participants during their visitation to our institution. Measurements of body height (BH), body weight (BW), systolic blood pressure (SBP) and diastolic blood pressure (DBP) were collected; BMI was calculated by dividing the were divided into the following groups according to age: subjects younger than 30 years, subjects aged between 30 to 60 years and subjects older than 60 years.

Statistical analysis
All statistical analyses were performed separately by gender and age groups with the software of Statistical Package for Social Sciences (SPSS version 17.0, Chicago, IL, USA). Statistically signi cance was de ned as P <0.05. Most of the factors could be described using a normal distribution except ALT, TB TG, HDL and G. Independent sample's t test was used to compare the differences between the indicators for normal distributed variables, and the results were expressed as mean ± standard deviation (SD).
Continuous variables with a skewed distribution, analyzed by Mann-Whitney U test, were described with median and interquartile range. Spearman bivariate correlations were performed between TB and other variables in different genders. The concentration of TB was divided into four quartiles. Chi-square test was used to compare the inter-group prevalence differences of high TC. At last, binary logistic regression models were performed to analyze the crude and adjusted odds ratios for high TC with 95% con dence intervals.

Characteristics of the participants in different genders
Males were younger than females. Most of the parameters had higher levels in males than in females, except TC, HDL and LDL. Differences also existed between high TC group and normal TC group ( Table 1). All of the parameters were higher in high TC group than in normal TC group in both genders except ALT, TG, HDL and G. In high TC group, females were older than males. Males had higher levels of all parameters than females while HDL and LDL were higher in females. In normal TC group, females were younger than males. The levels of most parameters were higher in males than in females. Only HDL was higher in females than in males. Incidence of high TC according to TB quartiles Incidence of high TC was compared between genders according to the quartiles of serum TB levels. Females showed signi cantly higher overall incidence of high TC than males in the second to the fourth TB quartiles. The incidence decreased in higher TB quartiles in both genders (Fig. 1A). Incidence of high TC was also compared between men and women of different age groups. For people younger than 30 years, men showed a signi cantly higher incidence of high TC than women. The incidence decreased with the increasing of TB quartiles in men, while women showed a decreasing trend in the second to the fourth TB quartiles. For people aged from 30 to 60 years, women had a signi cantly higher incidence of high TC than men only in the second and the fourth TB quartiles. For people older than 60 years, the incidence of high TC was higher in women than in men. There was almost no obvious decreasing of the incidence from the rst to the last TB quartiles in women. Men showed a mildly decreasing of the incidence from the rst to the last TB quartiles (Fig. 1B, 1C and 1D).

Risks of high TC in different genders
High TC risks for different genders were calculated by binary logistic regression models. All participants were analyzed in different genders (Table 2). Model 2 showed a protective effect of TB on high TC in both genders; for instance, the adjusted odd ratio (OR) of TB showed such a role (males: 0.985; females: 0.974). After adjusting for age, BMI, SBP, DBP, ALT, BUN, Cr, UA, TG, HDL, LDL and G, TB showed no association with TC both in women and men. This relationship was also not changed with the rising of TB quartiles in both genders. In addition, the appeal results were not changed in the binary logistic regression analysis of different age groups (Table 3).   TB = total bilirubin, OR = odds ratio, CI = con dence interval.
Logistic regression model 1 with TB Quartile 1 as reference, including age, body mass index, blood pressure, alanine aminotransferase, blood urea nitrogen, creatinine, uric acid, triglycerides, highdensity lipoprotein, low-density lipoprotein and glucose as covariates.
Model 2 and 3 analyzed TB as a continuous variable. Model 2 included TB as a covariate and model 3 included age, body mass index, blood pressure, alanine aminotransferase, blood urea nitrogen, creatinine, uric acid, triglycerides, high-density lipoprotein, low-density lipoprotein and glucose as covariates.
Logistic regression model 1 with TB Quartile 1 as reference, including age, body mass index, blood pressure, alanine aminotransferase, blood urea nitrogen, creatinine, uric acid, triglycerides, highdensity lipoprotein, low-density lipoprotein and glucose as covariates.
Model 2 and 3 analyzed TB as a continuous variable. Model 2 included TB as a covariate and model 3 included age, body mass index, blood pressure, alanine aminotransferase, blood urea nitrogen, creatinine, uric acid, triglycerides, high-density lipoprotein, low-density lipoprotein and glucose as covariates.

Discussion
Our study found no relationship between TB and TC in both genders of all age groups. Recent articles have described different oppinions. J.T. Hughes et al. [14] found that serum TB concentration was positively related with TC in 594 Indigenous Australians. Recent reports found the use of statins have a low negative effect on TB concentration [23]. So they thought this reverse causality could explain the positive correlation between TB and TC. On the contrary, negative relationship between TB and TC was also reported in several articles. Pearson correlation analysis suggested a signi cantly negative correlation between TC and TB in some researches [12,13], yet their Pearson correlation coe cients were in low value ranges, similar to our results. No further research was expended to prove their conclusions. Some articles reported that TB treatment decreased the levels of TC and TG in Gunn rats [24,25].
However, the results of animal studies are not completely applicable to human. In a cross-sectional analysis of 1,711 subjects with type 2 diabetes, TB was inversely associated with age, TC and TG levels in men [13]. The narrow participants ranging and the impact of diabetes make the results of the study less representative of healthy populations. Jenko-Praznikar, Z. et.al. [12] proposed a negative relationship between TB and TC in overweight asymptomatic middle-aged adults (64 women and 32 men). Another study on TB and cardiovascular disease found inverse correlations between TB and BMI, LDL and TC [11]. This study also lacked demographic representation (868 asymptomatic individuals). They speculated that other cardiovascular risk factors caused a negative correlation between TB and TC.
The negative relationship between TB and TC showed by our Spearman correlation study can be explained next. Firstly, it might be the effect of oxidative stress. Heme oxygenase catalyzes the cleavage of the heme ring, which forms ferrous iron, carbon monoxide and bilirubin. Biliverdin reductase (BVR) rapidly reduces biliverdin to bilirubin [26]. BVR physiologically regenerates bilirubin in the catalytic cycle, thereby providing the cytoprotective effect of anti-oxidants. Bilirubin had been reported to be an important chain-breaking anti-oxidant that prevents lipid peroxidation [27] and was also a physiological lipidlowering agent [28]. As an anti-oxidant, the level of TB in the body might be changed with its anti-oxidant effect. Therefore, the association between TB and TC might be caused by the occurrence of oxidative stress [12]. Animal studies could also con rm this speculation. In diabetes mellitus type I Sprague-Dawley rats models [29], serum TC and both serum and hepatic TG were decreased by TB. In the meanwhile, this effect accompanied by suppression of oxidative stress, as well as liver X receptor-α and Sterol regulatory element binding protein-1 which involved in regulating lipid synthesis and metabolism [30]. Secondly, insulin resistance could in uence their relationship. BVR was both a substrate for insulin receptor (IR) tyrosine kinase activity and an IR substrate 1 (IRS-1) serine phosphorylated kinase [31]. And IRS-1 serine phosphorylation was thought to be a mechanism of insulin resistance [32]. Therefore, the increased incidence of insulin resistance would be accompanied by an increase in BVR-catalyzed redox reactions, which would affect the level of TB. In a cross-sectional study [33], 12342 Korean adults aged 20 years and over were involved to investigate the negative relationship between TB and metabolic syndrome. Oxidative stress accompanied the development of metabolic syndrome which was associated with insulin resistance and dislipidemia [33]. They suggested that TB levels were in uenced by the insulin resistance-related dyslipidemic status, which could explain the inverse association between TB and TC.
Our study suggested that TB was not a risk predictor for high-TC in a large range of Chinese after excluding the confounding factors. There were no differences in both genders of different age groups.
Several studies were consistent with our results. A previous study has showed no relationship existed between TB and blood lipids in 327 heart failure patients with a preserved left ventricular ejection fraction [16]. Jaeseong Jo et. al. [17] proposed that direct bilirubin had a stronger relationship with metabolic syndrome than TB in Korean population. In a cross-sectional and longitudinal study of healthy population [34], the authors found no association between baseline TB and the incidence of dislipidemia in men. No causal relationship existed between TB and lipid status. A previous study [15] reported that the reduction of TB in diabetic patients was accompanied by increased levels of TG, LDL, TC and decreased levels of HDL. But this phenomenon was not found in the healthy control group. However, their sample size was too small (53 type 2 diabetes mellitus patients and 53 non-diabetic subjects) to meaningfully explore potential interactions. Our previous research also showed that TB could not be a risk factor of hypertriglyceridemia in a healthy Chinese population [35]. In summary, our results imply that there is no direct connection between TB and dyslipidemia, and TB is probably just a covariate in the development of dyslipidemia.
There are several limitations in our study. Firstly, it is a cross-sectional study, further longitudinal studies was needed to explore the causal relationship between TB and TC. Secondly, the direct bilirubin and indirect bilirubin were not obtained. Thirdly, failure to detect sex hormone levels in participants, which might affect the level of TB. Fourthly, it was di cult to identify corroborating details of other factors which conduce to the balance of anti-oxidant and oxidant stress among participants including poor eating habits, physical inactivity and a comprehensive assessment of tobacco.

Conclusions
In accordance with previous data on a general population, our ndings suggested that serum TB was not a risk factor of high TC in a healthy Chinese population. Longitudinal studies are necessary to determine whether serum TB levels have predictive value for high TC in both genders. The organizational review committee and the ethics committee of General Hospital A liated to Tianjin Medical University approved the ethics, methods and protocols of the survey. Signed informed consents were collected from all participants. We con rm that all methods were carried out in the light of the relevant guidelines and rules.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request

Competing interests
The authors declare that they have no competing interests. Incidence of high TC in different genders by TB quartiles. Picture (A) analyzed all participants, while pictures (B, C, D) separately analyzed participants in diferent age groups. Participants were divided into three groups according to age ( rst group: age ≤ 30; second group: 30 < age ≤ 60; third group: age > 60). For picture (A), TB quartiles 1 to 4 referred to the followings respectively: male: TB ≤ 9.9 umol/L, 9.9 umol/L < TB ≤12. , TB quartiles were also calculated separately in each gender with different age groups. * P < 0.05, ** P < 0.01.