Strengths and limitations of this study

The large sample size enhances the statistical power of the study.

The use of ultrasound for kidney stone (KS) diagnosis reduces the risk of misclassification bias.

The cross-sectional design limits the ability to infer causality.

Potential residual confounding may remain despite extensive adjustment for known risk factors.

The lack of data on KS type limits the analysis of subtype-specific associations.

Introduction

Kidney stones (KSs) have immense global health and economic burdens have been produced by KSs.1 A nationwide cross-sectional survey suggested that KS is common among Chinese adults, and one in 17 adults is affected by KS.2 Hence, identifying modifiable risk factors and developing effective strategies to prevent KS is urgently required. The prevalence of metabolic disorders such as insulin resistance, obesity and type 2 diabetes is dramatically increasing around the world,3 and previous evidence has suggested an association between metabolic abnormalities and stone formation.4 Among these metabolic factors, insulin resistance has garnered attention because of its role in promoting systemic inflammation and metabolic dysregulation, which are increasingly recognised as contributors to stone formation through various mechanisms, including alterations in the urinary composition and pH.5 The triglyceride-glucose (TyG) index, derived from fasting triglyceride and glucose levels, has emerged as a surrogate marker of insulin resistance. Elevated TyG levels have been found to be associated with various metabolic disorders, including obesity, type 2 diabetes and cardiovascular diseases.6

Evidence on the association between the TyG index and the risk of KS is still limited and inconsistent. Qin et al reported that a higher TyG index was associated with an increased likelihood of nephrolithiasis and recurrence in a nationally representative sample of adults from the National Health and Nutrition Examination Survey (NHANES) 2007–2014.7 Jiang et al reported that the TyG index was positively associated with KS prevalence in US adults from the NHANES 2007–2018.8 However, a recent longitudinal cohort study in a Taiwanese population suggested that the TyG index could not be used as a predictive factor for KS development.9 In these studies, KS was self-reported; hence, KS cases might have been underestimated, resulting in a misclassification bias. Moreover, the association between TyG index and KS remains unclear in Chinese adults.

In this case–control study, we aimed to investigate the association between TyG index and the risk of prevalent KS in a relatively large sample of Chinese adults.

MethodsStudy design and data collection

KS cases and non-KS controls were consecutively recruited from among those who underwent physical examinations at Ruijin Hospital, Pudong New Area People’s Hospital, and Shanghai Civil Aviation Hospital from January 2020 to December 2022. In total, 155 058 participants were screened for eligibility. 202 pregnant women and 302 patients with malignant tumours were excluded. In addition, 22 323 adults were excluded due to missing information on covariates and KS. Participants aged <20 or ≥80 years or with body mass index (BMI) <18.5 or ≥40.0 kg/m2 were further dropped out. Finally, 117 757 participants, including 11 645 patients with KS and 10 6112 KS-free controls, were included in the analysis (figure 1).

Figure 1

Flow of participants selection. BMI, body mass index.

Ascertainment of KS cases and TyG index

Healthcare professionals use ultrasound imaging to diagnose KS. KS was diagnosed on the basis of the presence of high echogenicity on ultrasonography, characterised by bright echoes with posterior acoustic shadowing. Overnight fasting blood samples were collected to measure the blood glucose and triglyceride levels. Biochemical analyses were conducted in a hospital laboratory using an automated analyser (Beckman AU5800). TyG index was calculated as ln[fasting triglyceride (mg/dL)×fasting glucose (mg/dL)/2].

Covariates

Demographic information such as age and sex was collected using questionnaires. Fasting blood samples were collected to measure serum creatinine and uric acid (UA) levels and lipid profiles (eg, total cholesterol (TC), high-density lipoprotein cholesterol (HDLC) and low-density lipoprotein cholesterol (LDLC)) using an automated analyser (Beckman AU5800). Anthropometric information was collected by trained health technicians and BMI was calculated as weight (kg) divided by height (m) squared. Subjects with BMI≥28.0 kg/m2 were considered obese. Hypertension was defined as self-reported physician-diagnosed hypertension or use of antihypertensive medication. Diabetes was defined as self-reported physician-diagnosed diabetes, hypoglycaemic drugs or insulin therapy. Estimated glomerular filtration rate (eGFR) was calculated using the formula proposed by Ma et al.10

Statistical methods

The characteristics of the study participants are summarised as mean (SD) or median (IQR) for continuous variables and numbers (percentages) for categorical variables. The differences in TyG index and demographic and clinical characteristics between individuals with and without KS were evaluated using Student’s t-test (parametric distribution) or Mann-Whitney U test (non-parametric distribution) for continuous variables and χ2 test for categorical variables. Moreover, the mean and SD of the TyG index across participant characteristics (age, sex, obesity, hypertension, diabetes and eGFR) were calculated. Student’s t-tests were used to test whether the TyG index varied across participant characteristics.

Participants were categorised into quartiles according to TyG level in the control group. We used a logistic regression model to examine the associations between the TyG index and the risk of prevalent KS, and calculated multivariable-adjusted ORs and 95% CIs across increasing quartiles. Two multivariate models were developed in this study. Model 1 was adjusted for age (years), sex (male and female) and BMI (kg/m2). In model 2, we further controlled hypertension (yes, no), diabetes (yes, no), LDLC (mmol/L), serum UA(μmol/L) and eGFR (mL/min/1.73 m2). The p value for the linear trend was calculated by introducing the medians of quartiles as continuous variables into the model. The multivariable-adjusted OR with 95% CI for KS associated with per SD increment in TyG index was also calculated. A restricted cubic spline (RCS) with five knots at the 5th, 27.5th, 50th, 72.5th and 95th percentiles of the TyG index distributions was plotted to examine the dose–response relationship between the TyG index and KS, and the reference value was set at the 10th percentile.

Stratified analyses by age (<50 and ≥50 years), sex (male or female), obesity (yes or no), hypertension (yes or no), diabetes (yes, no) and eGFR (<90, ≥90 mL/min/1.73 m2) were performed to examine whether these factors modified the association between TyG index and risk of KS. The potential interaction between the TyG index and stratification factor was evaluated by introducing a multiplicative term between the TyG index and stratification variable as continuous variables into the multivariate models and testing whether the coefficient of the interaction term was equal to zero.

Several sensitivity analyses were conducted to evaluate the robustness of the results. First, we applied propensity score matching (PSM), an increasingly popular method for adjusting for confounding, to cope with observed confounding.11 The propensity score (PS) for an individual is defined as the probability of KS given a set of covariates. Specifically, potential confounders, including age, sex, BMI, hypertension, diabetes, serum UA, LDLC and eGFR, were fitted to a multivariable logistic model. KS cases were matched with controls using 1:1 nearest neighbour matching with a calliper of 0.10 and exact matching of sex to avoid pairing dissimilar individuals. The standardised mean differences of all covariates in the matched and unmatched samples are plotted. We then applied a conditional logistic regression model to examine the association between TyG index and KS. Second, we excluded outliers of the TyG index (defined as <1st or >99th percentile) to examine whether our results were sensitive to influential observations. Finally, we calculated an assessment of potential residual confounding with E-values, defined as the minimum strength of association on the OR scale that an unmeasured confounder would need to have with both the exposure and outcome to fully explain the observed exposure-outcome association, conditional on the measured covariates.12 13

All statistical analyses were conducted using R V.4.3.2 (The R Foundation for Statistical Computing, Vienna, Austria). RCS was conducted with R package ‘plotRCS’ (V.0.1.5), PSM was conducted with R package ‘MatchIt’ (V.4.5.5) and E-values were calculated with R package ‘EValue’ (V.4.1.3). Two-sided p values <0.05 were considered statistically significant.

Patient and public involvement

None.

ResultsCharacteristics of the study population

A total of 11645 KS cases (mean age 53.25 years, 61.83% male) and 106 112 controls (mean age 55.46 years, 50.66% male) were included in this case–control study (table 1). Compared with control subjects, KS patients were younger and more likely to be male, obese and non-hypertensive; had higher values of BMI, fasting plasma glucose (FPG), TC, TG and serum UA; and lower values of HDLC, LDLC and eGFR (table 1). Means (SDs) of TyG index were 8.65 (0.63) for the control group and 8.74 (0.66) for the KS group. TyG values were significantly higher in individuals with KS than in the control participants (p<0.001). The means (SDs) of TyG according to the population characteristics are shown in online supplemental table S1. Older, male and obese adults had significantly higher TyG levels (all p<0.001). Moreover, significantly higher values of TyG were observed in those with hypertension, diabetes and eGFR<90 mL/min/1.73 m2 (all p<0.001).

Table 1

Characteristics of study participants

Associations between TyG index and risk of KS

Compared with the lowest quartile, the multivariable-adjusted ORs (95% CIs) for KS across increasing quartiles were 1.10 (1.03 to 1.16), 1.15 (1.08 to 1.22) and 1.28 (1.20 to 1.36), respectively (table 2). Moreover, each SD increment in TyG was associated with a 10% (OR: 1.10, 95% CI 1.08 to 1.13) greater risk of prevalent KS (table 2). RCS also suggested a significantly positive and linear association between TyG and KS (p overall<0.001 and p non-linear=0.186, respectively) (figure 2).

Figure 2

OR for kidney stone by TyG index. Line represents multivariable-adjusted OR, and shaded area represents 95% CI. Model was adjusted for age (years, continuous), sex (male, female) and BMI (kg/m2, continuous), hypertension (yes, no), diabetes (yes, no), LDLC (mmol/L, continuous), serum uric acid (μmol/L, continuous) and eGFR (mL/min/1.73m2, continuous). BMI, body mass index; eGFR, estimated glomerular filtration rate; LDLC, low-density lipoprotein cholesterol; PSM, propensity score matching; TyG, triglyceride-glucose.

Table 2

Association between TyG index and risk of kidney stone

To investigate whether confounders, including age, sex, obesity, hypertension, diabetes and eGFR, modified the association between TyG index and KS, we performed stratified and interaction analyses (online supplemental table S2). A positive association between TyG and KS persisted in all subgroups and evidence of effect modification by age, sex and hypertension was observed (online supplemental table S2). Compared with adults aged ≥50 years, each 1-unit increment in TyG was associated with a greater risk of KS in those aged <50 years (p interaction=0.02). In male adults, the multivariable-adjusted OR (95% CI) for KS associated with each 1-unit increment in TyG was 1.24 (1.19 to 1.29), which was significantly higher than that in female adults (p interaction<0.001). Moreover, a stronger positive association between TyG and KS was observed in patients without hypertension (p interaction=0.005).

Sensitivity analyses

PSM has become an increasingly popular method for adjusting for confounders in observational studies. 11 643 controls were matched with 11643 KS cases based on confounders (online supplemental table S3). After PSM, the cases and controls were well-matched, with no significant differences in all covariates (online supplemental table S3). Compared with the control subjects, patients with KS had significantly higher FPG, TC, TG and TyG values (all p<0.05) (online supplemental table S3). Standardised differences <0.10 for all covariates in PSM indicated a balance between KS cases and controls (online supplemental figure S1). Compared with the lowest quartile, subjects in the highest quartile were confronted with a 28% (OR: 1.28, 95% CI 1.19 to 1.38) greater risk of KS after PSM (online supplemental table S4). Moreover, the risk of prevalent KS increased as the TyG index increased (p overall<0.001 and p non-linear=0.291, respectively) (online supplemental figure S2). The positive association between TyG and KS persisted, even after excluding outliers (online supplemental table S5 and figure S3). The observed OR of 1.28 could be explained by an unmeasured confounder that was associated with both TyG and KS by an OR of at least 1.88 each, above and beyond the measured confounders (online supplemental table S6). E-value sensitivity analysis suggested that unmeasured confounding of considerable strength was needed to fully negate the observed positive association between TyG and KS.

Discussion

In this case–control study of 11 645 KS cases and 106 112 controls, we examined the association between the TyG index and KS in Chinese adults. Compared with the lowest quartile, participants in the highest TyG quartile had a 28% higher risk of KS. Each SD increment in TyG level was associated with a 10% greater risk of KS. The positive association between TyG index and KS persisted after several sensitivity analyses. The results of the stratified analysis revealed that age, sex and hypertension influenced the association between the TyG index and risk of KSs. The incidence of obesity, hyperuricaemia, metabolic syndrome and a progressive decline in renal function increases with age, all of which are closely linked to the risk of KS. In older individuals, these factors may attenuate the association between the TyG index and KS risk.5 14 The association between the TyG index and KS is stronger in men, likely because of the protective effect of oestrogen in women, which mitigates this association.15 16 Additionally, the stronger link between TyG index and KS risk in non-hypertensive patients may stem from the fact that hypertension, a known risk factor for KSs, reduces the influence of the TyG index in hypertensive individuals.17 18

The TyG index has been gradually accepted as a surrogate biomarker for measure insulin resistance.19 Currently, epidemiological evidence on the association between the TyG index and KS remains limited and controversial. Consistent with previous studies, our study found that TyG values in patients with KS were higher than those in non-KS controls. Qin et al investigated the association between TyG and KS in 20 972 non-pregnant adults from the NHANES 2007–2014 and reported that each unit increase in the TyG index was associated with a 12% increased risk of KSs in a fully adjusted model.7 Similarly, our study found that each SD increment in TyG was associated with a 10% greater risk of KS. Moreover, Qin et al observed a linear association between the TyG index and the risk of KS, which is consistent with our results that indicated a significantly positive and linear association between TyG and KS risk (p overall<0.001, p non-linearity=0.136). Jiang et al evaluated the TyG index for the occurrence of KS in the NHANES 2007–2018 and found a positive association between the TyG index and KS risk. Before PSM, a high TyG index was associated with a 14% increase in the risk. After PSM, the risk of KS remained elevated in the high-TyG group.8 We consistently observed a positive association between TyG and KS, before and after PSM. However, a previous study in a Taiwanese population revealed no significant association between TyG index and the risk of incident KS (OR: 1.25, 95% CI 0.92 to 1.71).9 However, this study used self-reported questionnaires to assess the incidence of KS, which may have led to an underestimation of actual cases, introducing a potential bias in the results.

The potential mechanisms linking the TyG index and KS risk remain unclear20 21; however, there are several possible explanations. Insulin resistance may initiate renal acid/base imbalance, inflammation and oxidative stress, ultimately interfering with stone-forming factors such as urinary pH, calcium, citrate and oxalate levels.22 23 Insulin resistance is thought to disrupt renal acid-base metabolism, leading to decreased urine pH and an elevated risk of urate calculi formation.24 Moreover, insulin resistance can increase the risk of calcium stone formation by decreasing urinary citrate excretion.25 Citrate prevents KS formation by binding to calcium, reducing urine concentration and inhibiting crystal formation.

Our study has several strengths. First, our study investigated the association between the TyG index and KS in a relatively large sample of Chinese adults. Second, KS cases were ascertained using ultrasonography, thus minimising misclassification bias. However, the limitations of this case–control study should be acknowledged. First, the case–control nature of our study did not allow us to infer causality between TyG index and KS. Second, our results should be interpreted with caution because of residual confounding factors. Although we adjusted for many confounders in the multivariable models, unobserved residual confounding factors (eg, smoking and diet) were inherent in this observational study. However, E-value sensitivity analysis suggested that unmeasured confounding of considerable strength would be needed to fully negate the observed positive association between TyG and KS. Third, no information was available regarding KS type.

Conclusions

A higher TyG index, a biomarker for insulin resistance, was associated with an increased risk of prevalent KS in a Chinese population-based case–control study. Our results provide novel insights into the prevention and management of KS. However, the results should be interpreted cautiously because of limitations such as the case–control design and residual confounding. Future longitudinal studies are warranted to validate our results, and experimental studies are required to gain a better understanding of the mechanisms linking insulin resistance and KS.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Consent obtained directly from patient(s).

Ethics approval

Ethical approval was obtained from the Ruijin Hospital (2024-177), Pudong New Area People’s Hospital (2024-LW-05) and Shanghai Civil Aviation Hospital (202208). Written and oral informed consent was obtained from all the participants prior to data collection, and their participation was voluntary. The research team assured the respondents that the data would be used solely for research purposes and handled them anonymously, with access restricted to the research team.

Acknowledgments

We would like to thank Mr Zhu Rui for his hard work in establishing and managing the database.