Introduction

Chronic kidney disease (CKD) has become a significant public health and economic burden, extensively affecting approximately 8% to 16% of the global population. The Global Burden of Disease Study 2017 have reported that there were approximately 697.5 million patients with CKD worldwide, which had caused 1.2 million deaths directly [1]. Additionally, the global mortality from CKD also shows a 41.5% increasing tendency [2]. People with CKD are characterized by chronic and progressive kidney dysfunction and were susceptible to progressing to end-stage renal disease (ESRD) [3]. In fact, the most important adverse outcomes of CKD include not only the complications of a decreased glomerular filtration rate (GFR) and progression to kidney failure but also an increased risk of cardiovascular disease (CVD) [4]. Although the exact pathological process of CKD remains unclear, but many studies have confirmed that several autoimmune diseases, genetic disorders, environmental pollution and increasing prevalence of diabetes and hypertension are critical risk factors for CKD [5,6,7,8].

Accumulating studies have indicated that insulin resistance (IR) might have a close relationship with CKD occurrence and progression. IR represents insensitivity to insulin in the human body and is very common in people with type 2 diabetes mellitus (T2DM) [9]. A prospective cohort study reported that there was a positive association between IR condition and CKD severity [10]. R A DeFronzo et al. pointed out that IR also existed in nondiabetic patients with ESRD [11]. Several studies demonstrated that severe IR partially contributed to the increase in cardiovascular mortality in patients with CKD [12, 13]. Kayo Shinohara et al. further concluded that IR could be used as an independent predictor of CVD mortality in patients with ESRD [14].

The triglyceride–glucose (TyG) index is a novel, plausible, and reliable indicator of IR based on fasting blood glucose and fasting triglycerides in clinical practice [15, 16]. The TyG index even performed better in assessing IR conditions than the traditional homeostasis model assessment of insulin resistance (HOMA-IR) and the triglyceride/HDL-cholesterol ratio [17, 18], partially because the distribution of body fat was heterogeneous in the whole human body [19]. Previous studies showed that a higher TyG index was associated with more severe metabolic syndrome, T2DM, liver fibrosis, and an increased risk of CVD [20,21,22,23]. However, the relationship between the TyG index and CKD remains unclear. Therefore, we utilized the publicly available database National Health and Nutrition Examination Survey (NHANES) cycle 2015–2016 and 2017–2018 to evaluate the TyG index levels in U.S. adults and explore its potential associations with kidney function indicators from a large-scale population perspective.

Materials and methods

Study population

We obtained data from NHANES 2015–2016 and 2017–2018 and enrolled a total of 4361 participants. NHANES is an ongoing cross-sectional and national survey managed by researchers in the National Center for Health Statistics (NCHS) at the U.S. Centers for Disease Control and Prevention (CDC). Samples in NHANES represented the health and nutritional status of the general U.S. population well and employed a carefully conducted multistage and stratified probability design. NHANES is publicly available at www.cdc.gov/nchs/nhanes/.

All participants ≥ 20 years old with full information on fasting glucose, serum triglyceride and creatinine, and urinary albumin creatinine ratio (UACR) from NHANES 2015–2018 were included in this study. NHANES 2015–2018 primarily included 19,225 participants, and we excluded participants below 20 years old (N = 7937), participants without serum triglyceride and fasting glucose levels to calculate the TyG index (N = 6473), participants without serum creatinine levels to calculate eGFR and participants who lacked UACR data (N = 454). The flow diagram of sample collection is shown in Fig. 1.

Fig. 1
figure 1

The flow diagram of sample selection

Full size image

Exposure and outcome definitions

We treated the TyG index as an exposure variable, and it was calculated by Ln (triglycerides (mg/dl) * fasting glucose (mg/dl)/2) [23]. According to the NHANES protocol, both triglycerides and fasting glucose were measured through enzymatic assays in an automatic biochemistry analyser. Serum triglyceride and fasting glucose were measured utilizing the Roche Modular P and Roche Cobas 6000 chemistry analysers and the hexokinase-mediated reaction on Roche/Hitachi Cobas C 501 chemistry analysers, respectively [24].

The outcome variables included eGFR, UACR, low eGFR, and albuminuria to serve as kidney function indicators. The CKD-EPI formula was used to calculate the eGFR (mL/min/1.73 m2 BSA), and "low eGFR" was defined as eGFR < 60 mL/min/1.73 m2 BSA [25, 26]. The measurement of UACR (mg/g) was clearly reported elsewhere, and albuminuria was defined as UACR > 30 mg/g [27, 28]. According to Kidney Disease: Improving Global Outcomes 2012 recommendations, CKD was defined as the existence of either eGFR < 60 mL/min/1.73 m2 BSA, or markers of kidney damage (e.g. albuminuria), or both, of at least 3 months duration, regardless of the underlying cause.[29]. CKD was categorized into 5 stages: CKD stage 1: eGFR ≥ 90 ml/min/1.73 m2 BSA; stage 2: 60 ml/min/1.73 m2 ≤ eGFR < 90 ml/min/1.73 m2 BSA; stage 3: 30 ml/min/1.73 m2 ≤ eGFR < 60 ml/min/1.73 m2 BSA; stage 4: 15 ml/min/1.73 m2 ≤ eGFR < 30 ml/min/1.73 m2 BSA; and stage 5: eGFR < 15 ml/min/1.73 m2 BSA [29].

Covariates

Continuous variables included age, energy and protein intake, systolic and diastolic pressure, glycohemoglobin, and serum levels of uric acid, phosphorus, total calcium and cholesterol. Categorical variables included sex, race, education level, marital status, ratio of family income to poverty (RIP), alcohol intake, body mass index (BMI), serum cotinine (smoking status), diabetes, and hypertension. RIP was classified into two groups (RIP > 1 or RIP ≤ 1) [30]. Alcohol intake was defined by 24 h dietary recall; if a participant reported any alcohol consumption, he or she would be grouped into "Yes", and vice versa. Participants were divided into three groups based on their BMI, including normal (BMI < 25 kg/m2), overweight (25 ≤ BMI ≤ 30 kg/m2) and obese (BMI > 30 kg/m2) [31]. We defined smoking exposure through serum cotinine levels: Smoker, serum cotinine > 10 ng/mg; Second-hand smoker, 0.011 ≤ serum cotinine ≤ 10 ng/mg; Nonsmoker, serum cotinine < 0.011 ng/mg [32]. Participants with a self-reported diabetes diagnosis, HbAlc ≥ 6.5%, a plasma glucose level ≥ 200 mg/dL at 2 h after oral glucose tolerance test (OGTT), a fasting glucose level ≥ 126 mg/dL, or use of oral hypoglycemic agents and/or insulin were all categorized into "diabetes" [30]. Hypertension was defined as a self-reported hypertension diagnosis, diastolic blood pressure ≥ 90 mmHg, and/or systolic blood pressure ≥ 140 mmHg [33].

Statistical analysis

We strictly followed CDC guidelines when conducting all statistical analyses, and a suitable sample weight for each participant was employed for the NHANES complex multistage cluster survey design [34]. Categorical variables are displayed as percentages, and continuous variables are presented as the mean ± standard deviation (S.D.). We assessed the difference among TyG tertiles through ANOVA for continuous variables and weighted chi-square tests for categorical variables. As for UACR (a non-parametric variable), we presented statistics in Median (P25, P75) and used Kruskal–Wallis H test for comparison. Multivariate linear regressions were employed to assess the association between the TyG index and UACR and eGFR, whereas multivariate logistic regressions were performed to evaluate the relationship between the TyG index and albuminuria and low eGFR. Three different models were assigned to explore the potential covariate impact on this association (Model 1: unadjusted; Model 2: adjusted for age, gender, and race; Model 3: adjusted for age, race, sex, energy intake, protein intake, systolic pressure, diastolic pressure, glycohemoglobin, serum uric acid, total calcium, serum phosphorus, total cholesterol, education level, RIP, marital status, alcohol intake, BMI, hypertension, diabetes, and serum cotinine). Interaction terms were also conducted to test the heterogeneity among different subgroups. All our analyses were performed by EmpowerStats software (www.empowerstats.com; X & Y solutions, Inc., Boston MA) and R version 4.0.5 (http://www.R-project.org, The R Foundation).

Results

Baseline characteristics of participants

Weighted baseline characteristics of participants grouped by TyG index tertiles are displayed in Table 1. A total of 4361 participants had an average age of 51 ± 18 years, of whom 53.34% were females and 46.66% were males. The mean TyG index was 8.60 ± 0.68, and the ranges of TyG index tertiles were 5.85 ~ 8.27, 8.27 ~ 8.85, and 8.85 ~ 12.39, respectively. Although the difference in energy intake among TyG index tertiles was not significant, people in tertile 3 tended to consume more protein (80.08 ± 46.64 g VS 75.82 ± 39.30 g, P = 0.0322) than those in TyG index tertile 1. Patients in TyG index tertile 3 had higher systolic and diastolic blood pressure, and the prevalence of hypertension was more serious in higher tertiles (Tertile 1: 10.60%; Tertile 2: 21.65%; Tertile 3: 27.06%, P < 0.0001). A significant increase in mean glycohemoglobin and a higher prevalence rate of diabetes were also observed in TyG tertile 3 (P < 0.0001). The mean ± SD eGFR was 92.80 ± 14.12 ml/min/1.73 m2 BSA and its difference among tertiles was not significant, but a dramatic increase in UACR (Tertile 1, 6.57(4.46, 29.49) mg/g; Tertile 2, 7.23(4.60, 12.98) mg/g; Tertile 3, 8.79(5.38, 19.80) mg/g, P < 0.0001) and a higher likelihood of albuminuria (Tertile 1, 8.33%; Tertile 2, 12.00%; Tertile 3, 17.85%, P < 0.0001) were identified among different TyG index tertiles. The prevalence rate of CKD in the general population was 13.35%, and it increased in line with a higher TyG index. Furthermore, people in the highest TyG index tertile were more likely to have a higher serum uric acid and to be males, overweight and obese, and smokers.

Table 1 Baseline characteristics of participants from 2015 to 2018 NHANES, grouped by TyG index tertiles
Full size table

Association between the TyG index and CKD

Associations between the TyG index and UACR and eGFR with a hierarchal covariate-adjusting model are shown in Table 2. Our results revealed a positive association between the TyG index and UACR (Model 1, β = 43.84, 95% CI: 29.84, 57.84, P < 0.0001; Model 2, β = 43.12, 95% CI: 29.10, 57.15, P < 0.0001; Model 3, β = 25.10, 95% CI: 6.76, 43.44, P = 0.0074), while that of eGFR was not significant. Then the TyG index was grouped into tertiles (sensitivity analysis), and we only found significant results in TyG index tertile 3 compared with the lowest TyG index tertile in model 1 (β = 55.06, 95% CI: 31.73, 78.39, P < 0.0001) and model 2 (β = 53.80, 95% CI: 30.44, 77.16, P < 0.0001). However, this association did not remain significant in the fully adjusted model (model 3). This result might indicate that the positive relationship between the TyG index and UACR could, at least, partially explained by extra covariates in model 3 in relative to model 2. No significant difference in eGFR among TyG index tertiles was observed.

Table 2 Associations between TyG index and kidney function in U.S. adults from NHANES 2015–2018
Full size table

We then investigated the association between the TyG index and the likelihood of albuminuria, low eGFR, and CKD in Table 3. The positive association between the TyG index and CKD was stable in three different models (Model 1, OR = 1.79, 95% CI: 1.59, 2.03, P < 0.0001; Model 2, OR = 1.79, 95% CI: 1.59, 2.02, P < 0.0001; Model 3, OR = 1.34, 95% CI: 1.13, 1.59, P = 0.0006). For model 3, the fully adjusted model, the OR reached 1.06 in tertile 2 (95% CI: 0.80, 1.42, P = 0.6750) and 1.41 in tertile 3 (95% CI: 1.05, 1.91, P = 0.0244) relative to TyG index tertile 1. This positive association was similar to the albuminuria prevalence. However, no marked correlation between TyG index and the prevalence rate of low eGFR was found.

Table 3 Associations between TyG index and low eGFR, albuminuria, and CKD in U.S. adults from NHANES 2015–2018
Full size table

Subgroup analysis

BMI-stratified subgroup analysis was then conducted to investigate the relationship between the TyG index and the likelihood of CKD, which is presented in Table 4. No statistically significant difference in the prevalence of albuminuria was found among the normal-weight group. However, the positive association between the TyG index and CKD became stronger in the overweight (OR = 1.61, 95% CI: 1.18, 2.20, P = 0.0027) and obese groups (OR = 2.48, 95% CI: 1.95, 3.15, P < 0.0001). In the obese group, the OR was 3.08 for participants in tertile 3, which was larger than the other groups (95% CI: 1.74, 5.43, P < 0.0001). However, there seemed to be no obvious significance between tertile 1 and tertile 2 in the obese group (P = 0.5432). Notably, there was no dependence on BMI categories for this association.

Table 4 Association between TyG index and CKD in U.S. adults from NHANES 2015–2018, stratified by BMI categories
Full size table

Table 5 displays the diabetes condition stratified association between the TyG index and the prevalence rate of CKD. Similarly, interaction terms (P for interaction = 0.4128) showed that diabetes condition did not affect this association. Interestingly, the TyG index was not correlated with the likelihood of CKD in people without diabetes in either the primary or sensitivity analysis. However, for people with diabetes, a higher TyG index was associated with a higher level of CKD (OR = 1.94, 95% CI: 1.46, 2.56, P < 0.0001). People in TyG index tertile 3 showed a 402% higher level of albuminuria (OR = 5.02, 95% CI: 1.58, 16.02, P = 0. 0.0063).

Table 5 Association between TyG index and CKD in U.S. adults from NHANES 2015–2018, stratified by diabetes
Full size table

Discussion

This cross-sectional study enrolled a total of 4361 participants from NHANES 2015–2016 and 2017–2018 and investigated the association between the TyG index and CKD. We identified that a higher TyG index was strongly associated with a higher UACR level and a higher level of CKD, especially in overweight and obese participants and/or individuals with diabetes. However, TyG index failed to predict the incidence of renal insufficiency, implying that the TyG index was more suitable for early prediction of CKD.

The TyG index has become an accessible and attractive option to assess IR condition due to the highly available and inexpensive biochemical markers needed for its calculation since it did not need quantification of insulin and might apply to all participants with different health condition [15, 16, 19, 35]. Previous studies showed that the TyG index even performed better in assessing IR conditions than the traditional IR indicators, such as homeostasis model assessment of insulin resistance (HOMA-IR) and the triglyceride/HDL-cholesterol ratio [17, 18]. Moreover, accumulating studies have identified that TyG index was not only a good indicator of IR condition but also had some specific superiorities in assessing the occurrence and progression of several diseases. Kun Zhang et al. independently used HOMA-IR and TyG index to assess the relationship between IR and male hypogonadism in China and concluded that the TyG index had a better predictability than HOMA-IR [36]. A systematic review pointed that the TyG index correlated with other IR prediction methods well and appears to be more superior in predicting IR risk and other cardiometabolic risk factors in children and adolescents [37]. Researches from Mexico conducted another systematic review to evaluate the diagnostic accuracy of the TyG index in assessing IR condition, where a moderate-to-low quality evidence about the usefulness of the TyG index as a marker of IR was identified [38]. In total, the TyG index performed well in assessing IR condition with different population settings, but its predictability of the occurrence of CKD have not been studied.

To our knowledge, this is the first cross-sectional study exploring the association between the TyG index and CKD. A higher TyG index indicated a more serious IR condition, which might lead to multiple unfavorable health outcomes. Two similar cross-sectional studies using the NHANES database reported that a higher TyG index was associated with more severe abdominal aortic calcification and an increased risk of subclinical myocardial injury in U.S. adults [24, 39]. For kidney health management, a study from the Austrian Vorarlberg Health Monitoring and Promotion Program (VHM&PP) identified that the TyG index appeared to be related to the occurrence of end-stage kidney disease (ESKD) and mediated almost 50% of the total association between BMI and ESKD in the general population [40]. Liangjing Lv et al. reported that the TyG index was a potential predictor for diabetic kidney disease in T2DM patients through a combined cross-sectional and longitudinal analysis [41].

Our results indicated that the TyG index was positively associated with UACR level and albuminuria (a kidney damage marker), while it did not correlate with eGFR level, indicating a possible relationship among insulin resistance (IR)-diabetes-albuminuria. And several studies have also confirmed that IR followed by hyperinsulinemia did exist in an early phase of CKD and was prior to clinically significant GFR decline [42,43,44]. IR was also found to be associated with early glomerular hyperfiltration and then led to late glomerular damage in the early stage of diabetic nephropathy [45]. Therefore, we posit that TyG index could be a very good early-stage kidney damage indicator, especially diabetic kidney damage. However, the exact mechanism behind the association between the TyG index and CKD was still not clear. Many previous studies have investigated and explored a close interaction between CKD and IR. On the one hand, the progression of CKD characterized by metabolic acidosis, intestinal dysbiosis, activation of chronic inflammatory pathways or production of uremia toxin substances might lead to IR. Thomas SS et al. revealed that the common inflammatory condition of CKD patients could facilitate and aggravate IR conditions [46]. Elevated muscle cell inflammatory cytokines and mass transmembrane glycoproteins (signal regulatory protein-α, SIRP-α) via the NF-\(\kappa\) B-dependent pathway in CKD patients were identified. SIRP-α reacted with insulin receptors and insulin receptor substrate-1 (IRS-1), whose overexpression could reduce insulin signal transduction, promote protein hydrolysis, and then lead to a more serious IR condition [46]. SIRP-α knockdown with siRNAs in skeletal muscle cells improved IR conditions with increased insulin sensitivity [46]. CKD also caused defects in signaling through the IRS/phosphatidylinositol 3-kinase (PI3-K)/Akt pathway and then led to insulin insensitivity and IR [47]. White adipose tissue (WAT) is recognized as an essential source of oxidative stress and inflammatory cytokines. Immune cell labeling experiments showed ascending expression of CD68, IL-6, SOCS-3 and oxidative stress genes in subcutaneous WAT in patients with CKD, indicating that CKD was also associated with the inflammation of adipose tissue, thereby promoting IR [48,49,50]. Furthermore, subgroup analysis showed that higher TyG index was associated with higher levels of CKD only in overweight/obese or diabetic participant instead of people with normal BMI or without diabetes, indicating that TyG index might be more utilizable in detecting early-stage kidney damage (albuminuria) among people who have already had abnormal metabolic condition (e.g., diabetes or obesity). Additionally, our results were consistent with previous studies that people with higher TyG index tended to have higher risks of obesity and diabetes, which were also closely linked with kidney damage, so there might be multidirectional relationship among TyG index (IR condition), obesity, diabetes, and kidney damage [51, 52].

Our studies possess multiple strengths. This is the first large-scale cross-sectional study based on NHANES data with full consideration of NHANES sample design and CDC analytical guidelines. We combined two continuous NHANES cycles to improve the sampled cohort and lend greater stability to the data estimates. Our results showed that a higher TyG index could be a good predictor of CKD occurrence, which might be highly useful among people in disadvantageous socioeconomic conditions with no availability for direct measurement of kidney function. Moreover, higher TyG index (more serious IR) was believed to be not only associated with poorer renal outcome, like ESKD, but also higher risks of other common complications of CKD, including metabolic syndrome (MetS) and CVD [53]. TyG index served as an accessible and practical IR indicators for early-stage kidney damage (albuminuria), and the early detection and treatment for CKD (asymptomatic until its later stages) has been an important way to prevent kidney dysfunction progression, associated complications and reduce the impact of CKD on public health resources, indicating its possible utility in large-scale kidney function screening. Furthermore, we reduced the use of self-report questionnaire data to avoid recall bias as much as possible. For instance, we defined smoking exposure through serum cotinine instead of smoking questionnaire. However, there were also some limitations in this study. First and foremost, we could not obtain the causal relationship between the TyG index and the risk of CKD due to the cross-sectional design, and the relationship among IR, metabolic disorders and kidney function indeed seemed to be multi-directional based on previous studies [42]. To verify their relationships, more-well designed longitudinal studies are needed. And as for TyG index itself, though it has been considered as a feasible IR assessment tool in many studies, TyG index is still novel and its comparison to hyperglycemic clamp (gold standard for IR; expensive and laborious) or other well established IR measurement like HOMA-IR (based on non-routinely collected fasting insulin) in predicting kidney function might need further investigation. Additionally, we could not clarify type I or type II diabetes and then manage correlated subgroup analysis due to a lack of related data. To consider the impact of protein intake on kidney dysfunction (albuminuria), we used 24 h dietary recall data, which means that recall bias is still inevitable. We could not consider all residual confounders due to unmeasured or unknown variables. Last, we only incorporated participants aged more than 20 years old, so more studies based on a wider population setting should be performed.

Conclusions

Higher TyG index was strongly associated with a higher UACR level and higher values of albuminuria and CKD, which might be useful in kidney function screening especially among people in disadvantageous socioeconomic conditions with no availability for direct measurement of kidney function. However, more well-designed studies are still needed to validate this relationship.