Association between serum total bilirubin and diabetic kidney disease in US diabetic patients

Background Bilirubin has been widely reported to be a protective factor against diabetic kidney disease (DKD) in Asian populations. However, few large-sample analyses have been conducted in American populations. This study aimed to investigate the association between serum total bilirubin (STB) level and DKD in a US diabetic cohort. Methods This cross-sectional study enrolled participants from the National Health and Nutrition Examination Survey (NHANES) 2003–2018. Univariate and multivariate logistic regression analyses were performed to assess the association between STB level and DKD. Three models were conducted to control the potential confounding factors. Subgroup analysis was carried out for further validation. Results Among the 5,355 participants, the median age [interquartile range (IQR)] was 62 [52–71] years; 2,836 (52.96%) were male, and 1,576 (29.43%) were diagnosed with DKD. In the entire cohort, no significant association between STB level and DKD was observed in any logistic regression models (p > 0.05). Subgroup analysis revealed that, in U.S. diabetic males, STB levels > 11.98 µmol/L were associated with a nearly 30% lower risk of DKD than STB levels ≤ 8.55 µmol/L. Additionally, a moderate STB level (8.56–11.98 μmol/L) was found associated with a nearly 25% lower risk of DKD in U.S. diabetic patients over 65 years old. Conclusion The association of STB level with DKD may depict differences across diverse populations, among which the impact of race, sex, and age requires thorough consideration and relevant inferences should be interpreted cautiously.


Introduction
As a serious complication of diabetes, diabetic kidney disease (DKD) is characterized by persistent albuminuria and progressive decline in glomerular filtration rate (GFR) (1)(2)(3).Ultimately, it can progress to end-stage kidney disease, in which kidney function is severely impaired and patients require dialysis or kidney transplantation to sustain life (1)(2)(3).Understanding the factors associated with DKD is crucial for managing its occurrence and enhancing the quality of life of patients (1)(2)(3).
The pathogenic mechanisms of DKD are complex, among which oxidative stress plays significant roles (4)(5)(6)(7).Briefly, abnormal glucose metabolism leads to mitochondrial dysfunction and excessive production of free radicals, which interact with the intracellular antioxidant defense system.The resulting oxidative stress can induce lipid peroxidation, DNA damage, and protein oxidation.These issues directly lead to the destruction of the glomerular filtration membrane, injury of renal tubular cells, and renal interstitial fibrosis.Moreover, oxidative stress also induces and exacerbates inflammation, further increasing the production of oxygen-based free radicals and thus creating a vicious cycle.Overall, hyperglycemia, mitochondrial dysfunction, and inflammation play significant roles in oxidative stress in DKD.These processes interact with each other, promoting the occurrence and progression of oxidative stress, ultimately leading to renal damage and the progression of DKD (4)(5)(6)(7).
As a metabolite of hemoglobin, bilirubin functions as an effective antioxidant, protecting cells against damage from oxidative stress through interactions with the aforementioned processes (8)(9)(10).Briefly, bilirubin mitigates cellular damage of reactive oxygen species (ROS) by binding to ROS.In addition, bilirubin regulates mitochondrial function; namely, it improves respiratory-chain function, increases ATP production, and enhances cellular energy supply (11,12).Moreover, bilirubin reduces inflammation by inhibiting the activation of inflammatory cells and the release of inflammatory mediators, achieved through the regulation of related signal pathways (11,13).
On another front, bilirubin level can be influenced by polymorphisms in related genes, including but not limited to UGT1A1, OATP2, HMOX1, BLVRA, and NOS3 (14-17).Among these genetic variations, some may also affect susceptibility to DKD and thus contribute to protective mechanisms.For instance, bilirubin has been shown to be related to the generation and function of endothelial nitric oxide synthase (eNOS), which is involved in the regulation of renal and glomerular hemodynamics (18).It has been reported that NOS3, the gene encoding eNOS, harbors genetic susceptibility to increased bilirubin levels (17).Additionally, NOS3 polymorphisms are associated with altered susceptibility to DKD, and thus NOS3 is a candidate target gene in the treatment of DKD (19).
In recent decades, numerous studies have reported the correlation of serum total bilirubin (STB) levels with DKD (20,21).Although a high STB level is currently considered a protective factor against DKD, several studies have presented conflicting conclusions (20)(21)(22)(23)(24).Moreover, previous research has predominantly focused on Asian populations, such as Chinese, Japanese, and Korean, and data regarding European and American populations are limited (20,21).In the studies involving Western populations by Targher G et al. (22,23), STB was found to be either uncorrelated or positively associated with DKD risk in diabetic patients, inconsistent with the conclusions of most Asian population studies.The potential reasons behind these inconsistent results remain unclear and might be related to differences among different populations, such as variations in genetic and lifestyle factors.
Therefore, we conducted a large-sample study to investigate the association of STB level with DKD in the American population, using data from the National Health and Nutrition Examination Survey (NHANES).

Data source and study population
NHANES is a national survey of the civilian noninstitutionalized population conducted by the National Center for Health Statistics (NCHS), employing a cross-sectional, multistage, stratified, subgroup probability sampling study design on a twoyear cycle.Further details on NHANES are available at https:// www.cdc.gov/nchs/nhanes/.The NHANES study was approved by the NCHS Ethics Review Board in the US.
This study involved a total of 80312 participants from eight NHANES cycles spanning from 2003 to 2018.The flowchart of participant selection is provided in Figure 1.Eventually, 5355 patients with diabetes were enrolled.The demographics, examination results, questionnaire data, and laboratory data of the participants were extracted.

Definition of diabetes and DKD
In accordance with the standards of the American Diabetes Association and previous research, the diagnosis of diabetes was established through (1) a previous diagnosis by medical professionals, (2) fasting plasma glucose level ≥ 7.0 mmol/L, (3) glycohemoglobin (HbA1c) value ≥ 6.5%, or (4) current treatment for diabetes (25).Participants with an albumin/creatinine ratio (ACR) ≥ 30 mg/mg were classified as DKD patients and those with an ACR < 30 mg/mg as non-DKD patients (1-3).

Assessment of STB
As the dependent variable of this study, STB level (mmol/L) was measured using the method based on the NHANES Laboratory Procedures.Briefly, in the presence of a solubilizing agent, STB is coupled with 3,5-dichlorophenyl diazonium in a strongly acidic medium, and the intensity of the red azo dye formed is directly proportional to the STB level and can be determined photometrically (546 nm).In the present study, STB level was categorized into Q1 (≤ 8.55 mmol/L), Q2 (8.56-11.98mmol/L), and Q3 (≥ 11.98 mmol/L) according to tertiles.
Among these covariates, race/ethnicity was categorized into the following five groups: Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other race-including multi-racial.Education level was classified as below high school, high school graduate/GED or equivalent, and college or above.BMI was categorized into normal (< 25.0 kg/m 2 ), overweight (25.0 to < 30.0 kg/m 2 ), and obese (≥ 30.0 kg/m 2 ).Additionally, WC values > 88 cm and > 102 cm for women and men, respectively, were used to define abdominal obesity.Hypertension was defined as systolic blood pressure (SBP) ≥ 130 mmHg and/or diastolic blood pressure (DBP) ≥ 80 mmHg after repeated examination by a physician.For smoking status, participants were divided into never-smokers (individuals who never smoked or smoked < 100 cigarettes in life) and smokers (having smoked ≥ 100 cigarettes in life).

Statistical analysis
To obtain nationally representative results, appropriate sample weights were applied to account for oversampling and nonresponse.Weighted analyses were performed following the guidelines of the NHANES.
Categorical variables were expressed as numbers (percentages).Continuous variables with a normal distribution were presented as mean ± standard deviation (SD), whereas those with an abnormal distribution were presented as median [Q1, Q3].Weighted chisquared tests or Kruskal-Wallis tests were employed to compare basic characteristics among the three STB groups, and weighted chisquared tests or Mann-Whitney U tests were employed to compare basic characteristics between DKD and Non-DKD groups.
Weighted univariate logistic regression model (model 1) and multivariable logistic regression models (models 2 and 3) were employed to calculate the odds ratio (OR) and 95% confidence interval (95% CI) for the association between STB level and DKD.Model 2 was adjusted for age, gender, race, education level, and PIR.Model 3 was adjusted for the factors included in model 2 and for BMI, WC, presence of hypertension, TC, HDL, HbA1c, ALB, ALT, AST, BUN, GGT, SUA, SCR, GLB, and smoking status.
To further determine the potential effect moderators and explore the association between STB level and DKD, subgroup analysis was performed according to age, sex, race/ethnicity, education level, PIR, BMI, WC, smoking status, and presence of hypertension.Three logistic regression models were constructed.
Statistical analysis was performed using R statistical software version 4.2 (R Project for Statistical Computing).A two-sided p < 0.05 was considered statistically significant.
Table 1 shows the baseline characteristics based on STB levels, divided into Q1, Q2, and Q3 according to tertiles.Age, sex, race/ ethnicity, PIR, BMI, WC, and levels of DBP, TC, HbA1c, HSA, ALT, AST, GGT, UA, SCR, and GLB exhibited significant differences among the various STB groups (p < 0.05).Participants in the Q3 group tended to be male and have a BMI < 30 kg/m 2 .Additionally, they tended to be older, with higher PIR, and higher levels of HSA, ALT, AST, GGT, UA, and SCR.Table 2 shows the baseline characteristics based on the DKD status.Overall, age, race/ethnicity, education level, PIR, presence of hypertension, and levels of HDL, HbA1c, HSA, ALT, BUN, GGT, UA, SCR, and GLB exhibited significant differences between the DKD and non-DKD participants (p < 0.05).Compared with the non-DKD participants, those with DKD tended to be older, male, smoker, have an education level below high school, a BMI < 25 kg/ m 2 , hypertension, and higher levels of HbA1c, BUN, GGT, UA, SCR, and GLB.The proportion of participants with DKD varied by race/ethnicity.In contrast, PIR, and levels of HDL, HSA, and ALT were higher in the non-DKD group than in the DKD group.

Association between STB level and DKD
Three logical regression models were established to analyze the relationship between STB level and DKD, and the effect value was expressed as OR and 95%CI (Table 3).However, no significant association between STB level and DKD was observed in any of the models.
For the validation of the outcomes, subgroup analysis was performed (Tables 4 and S1).When age was stratified at 65 years old, all models indicated that for participants over 65 years old, an STB level of Q2 (8.56-11.98mmol/L) was associated with a lower risk of DKD than Q1 (< 8.55 mmol/L).The results (OR [95%CI]) were as follows: 0.72 (0.54-0.97) for Model 1, 0.71 (0.53-0.93) for Model 2, and 0.67 (0.50-0.89) for Model 3.However, no difference was found for participants below 65 years old.

Discussion
In this study, we investigated the association between STB level and DKD in a US diabetic cohort from the NHANES 2003-2018.Contrary to the reports on Asian populations, we observed no significant association between STB level and the risk of DKD in the entire diabetic cohort.However, subgroup analyses revealed that compared with low STB levels, high STB levels are associated with a reduced risk of DKD in US diabetic males, whereas moderate levels are associated with a reduced DKD risk in diabetics over 65 years old.
In the past twenty years, a series of cross-sectional studies have consistently suggested an association between high STB levels and a reduced risk of DKD in patients with diabetes (26)(27)(28)(29)(30)(31)(32)(33)(34)(35).Concurrently, animal studies have been conducted to explore the mechanism whereby STB confers protection against DKD (36,37).However, there have been a few studies with conflicting results, suggesting no relationship between STB level and DKD (22,24), or even positing STB as a risk factor for DKD (23).
In a cross-sectional study using data from the NHANES 2001-2006, Targher G et al. (22) found that STB level is not correlated with GFR or albuminuria in patients with diabetes (n = 1253).As an updated study, here we used samples from NHANES 2003-2018, and our sample size was larger than that of the study by Targher Considering our findings alongside the aforementioned research results, we propose that the association of STB level with DKD may exhibit distinctions among diverse populations.We speculate that the contrasting findings between Asian and Western populations can be attributed to several factors.Firstly, genetic variations may contribute to differences in both bilirubin metabolism and DKD susceptibility.For example, the NOS3 gene is implicated in the genetic susceptibility to elevated STB levels (17), and polymorphisms in this gene have been associated with altered DKD susceptibility (19).Although not conclusively confirmed, it cannot be ruled out that these genetic variations may influence the association between STB level and DKD.Secondly, lifestyle factors might also contribute to the observed differences.Western populations often have distinct dietary patterns, physical activity levels, and environmental exposures compared to Asian populations.These lifestyle factors may interact with STB levels and influence the development of DKD.Lastly, other factors that vary between Asian and American populations, such as comorbidities and socioeconomic status, may also confound the relationship between STB level and DKD.Further studies are needed to clarify the mechanisms and provide insights into this disparity.
In our further analysis, we found that the association between STB level and DKD varied with age and sex.Specifically, for US diabetic individuals over 65 years old, having STB levels in Q2 (8.56-11.98mmol/L) was associated with a nearly 30% lower risk of developing DKD than those with STB levels < 8.55 mmol/L.However, this association became statistically insignificant for individuals aged > 65 years with STB levels > 11.98 mmol/L, as well as for those aged < 65 years old, regardless of STB levels.Additionally, we found that males and females differed in the correlation between STB levels and DKD.Notably, a higher STB level (> 11.98 mmol/L) was observed to be associated with a nearly 25% reduced risk of DKD in males, whereas no significant correlation was found in females.Importantly, such sex-and agerelated differences in both STB level and DKD prevalence have been widely reported (38)(39)(40), as also reflected in our results (Tables 1  and 2).
Overall, we recommend thorough consideration of the impact of race, sex, and age when exploring the correlation between STB  The major strength of our study lies in the substantial sample size, as all the eligible participants from eight NHANES cycles (2003-2018) were included, ensuring the reliability of our conclusions.However, our study has some limitations.Firstly, despite revealing the association between STB level and DKD in specific subgroups, establishing causation is not feasible due to the crosssectional nature of the study.Prospective studies are necessary in the future to elucidate the causal relationship between STB level and DKD risk.Secondly, although we controlled for certain confounders in the subgroup analysis, there may still be unaccounted-for confounders not included or recorded in the NHANES.Thirdly, our study could not explore potential hereditary factors that might influence the association between STB level and DKD.Bilirubin levels exhibit significant genetic predispositions, such as hereditary hyperbilirubinemias.Additionally, hereditary susceptibility plays a role in the pathogenesis of DKD (41).Therefore, hereditary factors might influence the association between STB level and DKD.However, it is not possible to assess the family relationships among the participants and obtain information on hereditary factors in the data from the NHANES because it is a randomly sampled cross-sectional survey.We anticipate that prospective studies in the future will provide further elucidation and confirmation of the relationship between STB level and DKD.

Conclusion
The correlation of STB level with DKD could exhibit distinctions across diverse populations, among which the impact of race, sex, and age demands thorough consideration.Compared with STB levels < 8.55 mmol/L, STB levels > 11.98 mmol/L are associated with a nearly 30% reduced risk of DKD in US diabetic males.Furthermore, a moderate STB level (8.56-11.98mmol/L) is linked to a nearly 25% reduced risk of DKD in U.S. diabetic patients aged > 65 years.Further validation through large-scale prospective cohorts is warranted in the future.

FIGURE 1 Flowchart
FIGURE 1Flowchart of participant selection.
G et al. (n = 5355).Our analyses on this larger and updated sample set reproduced the findings of Targher G et al., thereby reinforcing the reliability of their conclusion.Another retrospective analysis by Targher G. et al. (23) indicated a negative correlation between STB and GFR levels in an unselected outpatient cohort of Caucasian non-diabetic and diabetic adults.This finding contrasts with conclusions drawn from the majority of previous studies based on Asian populations.

TABLE 1
Basic characteristics of the participants based on STB tertiles.

TABLE 2
Basic characteristics of the participants based on DKD.

TABLE 3
Association between STB and DKD. and DKD, and caution in interpreting relevant conclusions.Notably, this study represents the first large-scale investigation into the relationship between STB level and DKD in the US population.