Introduction

Central obesity is known to be independently associated with the risk of cardiometabolic diseases, and it is considered to be a more accurate cardiovascular risk predictor than overall obesity [1, 2]. Furthermore, in individuals with coronary artery disease (CAD), central obesity rather than body mass index (BMI) has been directly associated with mortality [3]. Central obesity has also been significantly associated with an elevated risk of cardiovascular disease (CVD)-related and all-cause mortality among adults within the normal weight range, particularly mortality related to heart diseases [4]. Central obesity, also termed visceral or abdominal obesity, is characterized by excess fat accumulation in the abdomen, and it has been associated with various health risks [5]. The prevalence of central obesity continues to increase worldwide [6], particularly in developed countries such as Germany, Spain, and the US, with reported rates of 33.9%, 36%, and 56%, respectively [7,8,9]. In Asia, the China Health and Nutrition Survey revealed marked increases in prevalence rates of overweight, central and general obesity from 1993 to 2015, especially among men [10]. In Taiwan, the prevalence of central obesity almost doubled from 1993 to 2016 [11]. Consequently, studies on risk factors for central obesity in patients with stable CAD are warranted.

Major risk factors for central obesity include genetic alterations [12], high-fat diet [13], physical inactivity [14], and sleep disturbance [15]. Other factors associated with a higher risk of central obesity include smoking and smoking cessation [16], low education and socioeconomic status [17, 18], depression [19], and stress [20]. In addition, a previous study demonstrated the key role of inflammation in the weight gain process [21]. Individuals with obesity have been shown to have a pro-inflammatory gut microbiota, which in turn has been shown to release uremic toxins, consequently causing inflammation and oxidative stress [22]. Total p-cresylsulfate (PCS) and indoxyl sulfate (IS) are harmful uremic toxins that have been shown to exert various detrimental effects, including the stimulation of fibrosis, oxidative stress, and inflammatory responses [23, 24]. Increasing evidence has suggested an association between high plasma levels of total PCS and IS with the progression to end-stage renal disease and a higher risk of mortality in patients with chronic kidney disease (CKD) [25, 26]. Furthermore, our previous studies found that elevated serum total PCS may play a role in the pathogenesis of coronary atherosclerosis [27] and left ventricular systolic dysfunction [28]. In addition, Koppe et al. suggested that total PCS may contributes to insulin resistance [29]. Despite these findings, little is known about the association between total PCS and IS levels with central obesity. A previous study demonstrated that certain central obesity anthropometric indices including a body shape index (ABSI) and conicity index (CI) may be more appropriate prognostic markers than BMI [30,31,32]. Therefore, the objective of this study was to explore associations between total PCS and IS with central obesity, as assessed by the CI and ABSI, in patients with stable CAD. In addition, as the potential mechanisms linking total PCS and IS to central obesity remain unclear, we selected six proteins related to obesity, and used structural equation modeling (SEM) to evaluate indirect and direct effects of total PCS, IS, and these six obesity-related proteins on central obesity. To address these objectives, we propose the following hypotheses: (1) Higher plasma levels of total PCS and IS are associated with increased central obesity, as measured by CI and ABSI, in patients with stable CAD. (2) The association between total PCS and IS levels with central obesity is mediated by specific obesity-related proteins. (3) The identified obesity-related proteins have both direct and indirect effects on central obesity through their interaction with total PCS and IS levels.

Methods

Ethics approval and consent to participate

In this cross-sectional study, all methods were performed in accordance with relevant guidelines and regulations. Additionally, the Human Research Ethics Committee of Kaohsiung E-Da Hospital approved the study (EDAH IRB No. EMRP-105-053). All participants provided written informed consent before enrollment.

Study population

We prospectively enrolled 399 consecutive hospitalized patients with stable CAD from the Cardiovascular Ward of our hospital from January 2022 to December 2022. In this study, we defined stable CAD as resting chest pain or pain related to effort which had remained stable for the preceding 6-month period. Patients were included into this study if they: (1) were >18 years of age when diagnosed with stable CAD; (2) had undergone cardiac catheterization to confirm the diagnosis of CAD; (3) did not have any missing clinical/medical data; and (4) could provide written informed consent. Patients were excluded if they: (1) were pregnant; (2) had other diseases related to inflammation (collagen and liver diseases, sepsis, infections, and malignancies); (3) had received steroid therapy or undergone surgery in the 1 month preceding admission; (4) lacked complete medical history/demographic data; and (5) could not provide written informed consent. After screening for the inclusion and exclusion criteria, a total of 373 patients were included in the study (26 were excluded).

In this study, the investigator was blinded to group allocation during the experiment and when assessing the outcomes. The extent of blinding included the following measures: (1) The allocation of samples to experimental groups was performed by a separate team member who did not participate in the data collection or analysis. (2) The investigator who assessed the outcomes did not have access to the group assignments and was provided with anonymized data. In addition, we conducted sex-stratified analyses to investigate potential differences in the associations between serum levels of total PCS, IS, and central obesity indices (CI, ABSI) in the male and female patients. This approach was used to account for potential sex-specific physiological and metabolic differences that could influence these associations. The sex-stratified analyses involved separate correlation analyses for male and female subgroups. The rationale for this stratification was based on previous literature suggesting distinct metabolic and hormonal profiles between sexes [33, 34] which may have impacted the relationships between PCS, IS, and central obesity indices. Details of the statistical procedures for the sex-stratified analyses are provided in the Statistical Analysis and Results sections.

Sample size calculation

To ensure the study had adequate power to detect significant associations between serum levels of total PCS, IS, and central obesity indices, we performed an a priori sample size calculation. The calculation was based on detecting a moderate effect size (Cohen’s d = 0.5) with a power of 80% (β = 0.20) and a significance level of 0.05 (α = 0.05). Using these parameters, the results showed a required sample size of 51 per group.

Power analysis

The final sample size of 373 patients included 257 males and 116 females. A posteriori power analysis was conducted to confirm the power for detecting the main associations in male and female subgroups. The analysis focused on the primary associations between total PCS, IS, and central obesity indices (CI and ABSI) in male and female patients, and indicated that the study had 99.9% power for males and 91.1% power for females to detect a moderate effect size, thereby minimizing the risk of type II errors and ensuring the reliability of our findings.

Collection of data

All of the included patients were ethnically Han Chinese and lived in the south of the country. The recorded sociodemographic data included sex, age, weight, height, and BMI (kg/m2). Detailed interviews were conducted using standardized questionnaires to gather information on personal medical histories, smoking, and alcohol consumption. In this study, we compared current and former smokers/drinkers (defined as having stopped smoking and drinking for 1 year or more) with individuals who had never smoked or drank alcohol. Height and weight were measured to the nearest 0.5 cm and 0.1 kg using standard procedures. Data on waist (measured at the narrowest point between the uppermost right iliac crest lateral border and lowest rib) and hip (measured at the widest point) circumferences were obtained with an accuracy of 0.1 cm, with each value being measured in duplicate. The waist-to-hip ratio (WHR) was subsequently calculated. CI was calculated as: CI = (waist circumference (m))/ (0.109 × √((body weight(kg))/ (height(m)))) [35], and ABSI was calculated as: ABSI = waist circumference (m)/[(BMI)^(2/3) × (height (m))^(1/2)] [31]. Blood pressure measurements were made by a trained nurse with the patients seated using a blood pressure monitor (HEM-907; Omron, Japan) after 5 min of rest.

Laboratory data

Complete blood cell count, serum uric acid, albumin, triglycerides, total, high- and low-density lipoprotein cholesterol (TC/HDL-C/LDL-C), and glucose were measured using 8-hour fasting blood samples [36, 37]. The concentration of serum creatinine was estimated using Jaffe’s technique, and hemoglobin A1c was assessed using glycohemoglobin analysis (HLC-723G8, Tosoh Corp., Tokyo, Japan). Renal function was assessed according to estimated glomerular filtration rate (eGFR), which was calculated with the CKD-EPI equation [38]. Peripheral leukocyte count was measured using an automated hematology system (XE-2100 Hematology Alpha Transportation System, Sysmex Corporation, Kobe, Japan).

Measurements of total PCS, IS, fibroblast growth factor 21 (FGF21), growth differentiation factor 15 (GDF15), fatty acid-binding protein 1 (FABP1), adiponectin, leptin, and high sensitivity C-reactive protein (hs-CRP)

As in our previous reports, we used ultra performance liquid chromatography (UPLC) to determine blood concentrations of total PCS and IS [28, 39]. Briefly, blood samples were centrifuged immediately after being drawn and kept at -80°C until assay. Deproteinization of the serum samples was achieved with the addition of three parts methanol to one part serum. A UPLC assay was used with a 2.1 × 100 mm ACQULITY UPLC bridged ethyl hybrid phenyl column to evaluate the concentrations of total PCS and IS at room temperature with PDA detection at 260 nm and 280 nm, limits of detection of 1 mg/L and 0.225 mg/L, and intra-/inter-assay coefficients of variation (CVs) of 5.50%/7.48% and 0.4%/0.05%, respectively, Calibration curves of peak area against the concentration of each analyte were plotted, with an average r2 value of 0.999 ± 0.001. Quantitative results are presented as concentration (mg/L). Concentrations of plasma GDF15 were determined with enzyme-linked immunosorbent assay (ELISA) kits (Quantikine Human GDF15 Immunoassay; R&D Systems, Minneapolis, MN), which had inter-/intra-assay CVs of 4.7–6.0% (n = 3) and 1.8–2.8% (n = 3), respectively. Plasma concentrations of FABP1 and FGF21 were also determined using ELISA (Cloud-Clone Corp., Katy, TX), with an analytical sensitivity of 0.59 ng/mL and intra-/inter-assay CVs of < 10% (n = 3) and < 12% (n = 3), respectively. Plasma concentrations of high molecular weight adiponectin and leptin were determined using ELISA kits (B-Bridge International, Sunnyvale, CA; and Quantikine Human Leptin Immunoassay; R&D Systems, Minneapolis, MN, respectively), with inter-/intra-assay CVs of 3.2–7.3% (n = 3) and 3.1-6.2% (n = 4), and 3.5–5.4% (n = 3) and 3.0–3.3% (n = 3), respectively. Plasma hs-CRP concentration was measured with a chemical analyzer (IMMAGE, Beckman Colter, Brea, CA), with a detection limit of 0.2 mg/L and intra-assay CV of 4.2–8.7%. We conducted each measurement twice per experiment.

Definitions

The definition of hyperlipidemia followed the ATP III criteria as the presence of one of the following (all values in mg/dL): (1) TC concentration of 200 or higher, (2) LDL-C concentration of 130 or higher, (3) triglyceride concentration of 150 or higher, (4) HDL-C concentration of less than 35/39 in men/women, (5) receiving medication for a lipid disorder [40]. Patients receiving medication for hypertension and those who had a systolic/diastolic blood pressure (SBP/DBP) of 140/90 mmHg or higher were defined as having hypertension. Similarly, patients receiving medication for diabetes mellitus (DM) and those who had a fasting glucose level higher than 126 mg/dL [41] were defined as having DM.

Statistical analysis

The Kolmogorov-Smirnov test was used to examine the normality of data. Continuous data with normal distribution were compared using the unpaired Student’s t-test and presented as mean ± SD, and those with non-normal distribution were presented as median (interquartile range). Categorical variables were compared using the chi-square test and presented as number (%). Logarithmically transformed values of triglycerides, hs-CRP, GDF15, FABP1, FGF21, adiponectin, leptin, total PCS, and IS were used due to their skewed distribution. Relationships among total PCS, IS, and biochemical and clinical variables were examined using Pearson’s correlation test, stratified by male and female subgroups. We defined body shape phenotypes using dichotomized CI and ABSI. Odds ratios (ORs) and 95% confidence intervals for the risk of CI ≥ 1.303 and ABSI ≥ 0.084 in each tertile of total PCS and IS were estimated with uni- and multivariate logistic regression analyses, with low total PCS and IS as the reference. Linear trends in risk were examined using one tertile as a continuous variable in the regression analysis. JMP version 10.0 (SAS Institute, Cary, NC) was used for the analysis, and G*Power was used for the power analysis. In addition, IBM SPSS AMOS version 24 (Amos Development Corporation, Meadville, PA) was used to fit the path model and SEM as shown in Fig. 3, with standard criteria including standardized root mean square residual < 0.06, root mean square error of approximation < 0.08, and comparative fit index > 0.90 as indices of the statistical fit of the models to the data [42]. In addition, model fit was estimated using the maximum likelihood method, and the results were presented as standardized path coefficients along with their statistical significance. A p-value < 0.05 was considered statistically significant.

Results

Characteristics of the male and female patients

The mean (± SD) age of the patients was 70.8 ± 12.3 years (range 35–99 years), 257 were male (68.9%) and 116 were female (31.1%) (Supplementary Table 1). The male patients were younger, and had higher hyperlipidemia, current smoker, and alcohol consumption rates, and higher WHR, uric acid, albumin, hematocrit, and hemoglobin than the female patients. In addition, the male patients had lower HDL-C, eGFR, adiponectin, and leptin levels than the female patients. There were no significant differences in DM, hypertension, BMI, SBP, DBP, fasting glucose, HbA1c, TC, triglycerides, LDL-C, creatinine, hs-CRP, GDF15, FABP1, FGF21, white blood cell (WBC) count, total PCS, IS, CI, and ABSI between the male and female patients (Supplementary Table 1).

Relationships of total PCS and IS with other parameters in the male and female patients

In the male patients, positive correlations were found between serum total PCS level with age, WHR, DM, hypertension, SBP, HbA1c, uric acid, creatinine, IS, CI, and ABSI, and negative correlations were found between serum total PCS level with BMI, hyperlipidemia, albumin, eGFR, hematocrit, and hemoglobin. Furthermore, we found positive correlations between IS with age, WHR, hypertension, SBP, uric acid, creatinine, CI, and ABSI, and negative correlations with hyperlipidemia, LDL-C, albumin, eGFR, hematocrit, and hemoglobin. In addition, in the female patients, we found positive correlations between serum total PCS level with DM, hypertension, HbA1c, creatinine, and IS, and negative correlations with BMI, albumin, eGFR, hematocrit, and hemoglobin. Furthermore, we found positive correlations between IS with DM, uric acid, and creatinine, and negative correlations with BMI, albumin, eGFR, hematocrit, and hemoglobin (Table 1). Moreover, simple linear regression analysis showed significant associations between the concentrations of serum total PCS and CI (β = 0.273, p < 0.0001, Fig. 1A) and ABSI (β = 0.297, p < 0.0001, Fig. 1B). The IS concentration was positively associated with CI (β = 0.260, p < 0.0001, Fig. 1C) and ABSI (β = 0.285, p < 0.0001, Fig. 1D) in the male patients, while no such association was found in the female patients.

Table 1 Relationship between total p-cresylsulfate, indoxyl sulfate and other parameters in males and females patients.
Full size table
Fig. 1: Association of log-transformed serum total p-cresylsulfate (PCS) and indoxyl sulfate (IS) concentrations with conicity index (CI) and a body shape index (ABSI) in male patients.
figure 1

The association between log-transformed serum total PCS concentration and CI and ABSI showed that log-transformed serum total PCS concentration had a significant positive association with CI A and ABSI B in male patients. In addition, the association between log-transformed serum IS concentration and CI and ABSI showed that log-transformed serum IS concentration had a significant positive association with CI C and ABSI D in male patients.

Full size image

Associations among total PCS, IS, CI, and ABSI status in the male patients

Although no associations were found between the concentrations of serum total PCS and IS with CI and ABSI in the female patients, the power analysis indicated that this study had sufficient power (91.1%) to identify such associations in the female patients. Therefore, we conducted further analysis of the male patients to examine whether total PCS and IS concentrations were associated with central obesity.

In this study, we used the median values of the whole sample as the cutoff points for CI and ABSI at 1.303 and 0.084, respectively. These cutoff points are consistent with those of previous studies [43, 44]. The percentage of patients with CI ≥ 1.303 and ABSI ≥ 0.084 increased along with the tertile of total PCS (Table 2). Accordingly, the ORs for the association with CI ≥ 1.303 increased with increasing PCS level relative to a low total PCS level (OR = 1.0; medium total PCS level, OR = 1.32; high total PCS level, OR = 3.61; p for trend across total PCS tertiles < 0.0001). Furthermore, higher ORs were found for the association with ABSI ≥ 0.084 relative to a low total PCS level (OR = 1.0; medium total PCS level, OR = 1.59; high total PCS level, OR = 3.90; p for trend across total PCS tertiles < 0.0001).

Table 2 Odds ratio (ORs) of conicity index and a body shape index status according to tertiles of total p-cresylsulfate in male patients.
Full size table

In addition, higher ORs were found for the association with CI ≥ 1.303 relative to a low IS level (OR = 1.0; medium IS level, OR = 1.66; high IS level, OR = 2.84; p for trend across IS tertiles = 0.001). Moreover, higher ORs were also found for the association with ABSI ≥ 0.084 relative to a low IS level (OR = 1.0; medium IS level, OR = 1.66; high IS level, OR = 2.84; p for trend across IS tertiles = 0.001; Table 3).

Table 3 Odds ratio (ORs) of conicity index and a body shape index status according to tertiles of indoxyl sulfate in male patients.
Full size table

Table 4 shows the results of uni- and multivariate logistic regression analyses for the associations between total PCS with CI and ABSI status. The patients with high total PCS levels had higher rates of CI ≥ 1.303 than those with low total PCS levels in all three logistic regression models (OR: 3.61, 95% confidence interval 1.95–6.88, p < 0.0001, OR: 3.67, 95% confidence interval 1.97–7.02, p < 0.0001, and OR: 3.18, 95% confidence interval 1.54–6.75, p = 0.002, respectively). In addition, the patients with high total PCS levels had a higher rate of ABSI ≥ 0.084 than those with low total PCS levels in all three models (OR: 3.90, 95% confidence interval 2.10–7.45, p < 0.0001, OR: 3.89, 95% confidence interval 2.07–7.48, p < 0.0001, and OR: 3.28, 95% confidence interval 1.54–7.24, p = 0.002, respectively). However, the patients with medium total PCS levels did not have higher rates of CI ≥ 1.303 and ABSI ≥ 0.084 than those with low total PCS levels in the three models.

Table 4 Logistic regression of the association of total p-cresylsulfate levels with conicity index and a body shape index status in male patients.
Full size table

Three univariate and multivariate logistic regression models were also used to examine associations between IS with CI and ABSI status (Table 5). The patients with high IS levels had a higher rate of CI ≥ 1.303 than those with low IS levels in the three models (OR: 2.84, 95% confidence interval 1.56–5.26, p = 0.001, OR: 2.72, 95% confidence interval 1.49–5.06, p = 0.001, and OR: 2.30, 95% confidence interval 1.15–4.66, p = 0.018, respectively). Furthermore, the patients with high IS levels had a higher rate of ABSI ≥ 0.084 than those with low IS levels in the three models (OR: 2.84, 95% confidence interval 1.56–5.26, p = 0.001, OR: 2.02, 95% confidence interval 1.06–3.88, p = 0.034, and OR: 2.22, 95% confidence interval 1.07–4.72, p = 0.033, respectively). However, the patients with medium IS levels did not have a higher risk of CI ≥ 1.303 and ABSI ≥ 0.084 compared to those with low IS levels in all three models, except for Model 3, where patients with medium IS levels had a higher rate of ABSI ≥ 0.084 than those with low IS levels (OR: 2.47, 95% confidence interval 1.15–5.44, p = 0.020). Furthermore, in the fully adjusted model, the patients with high total PCS levels had a 3.18-fold higher risk of CI ≥ 1.303 (Fig. 2A) and a 3.28-fold higher risk of ABSI ≥ 0.084 (Fig. 2B) compared to those with low PCS levels. Moreover, in the fully adjusted model, the patients with high IS levels had a 2.30-fold higher risk of CI ≥ 1.303 (Fig. 2C). In addition, the patients with medium and high IS levels had 2.47-fold and 2.22-fold higher risks of ABSI ≥ 0.084 (Fig. 2D), respectively, compared to those with low IS levels.

Table 5 Logistic regression of the association of indoxyl sulfate levels with conicity index and a body shape index status in male patients.
Full size table
Fig. 2: Odds ratios (ORs) of total p-cresylsulfate (PCS) and indoxyl sulfate (IS) levels with conicity index (CI) and a body shape index (ABSI) in male patients.
figure 2

ORs diagrams between total PCS - levels and a CI ≥ 1.303 A, ABSI ≥ 0.084 B, and ORs diagrams between IS levels and CI ≥ 1.303 C, ABSI ≥ 0.084 D, after adjusting for multiple confounders in male patients.

Full size image

Correlation matrix between CI, ABSI, and biochemical measures in the male patients

To explore potential causal interrelationships between total PCS, IS, and central obesity, we selected six proteins associated with obesity. We then used Pearson’s correlation analysis to assess the relationships between CI and ABSI and the six obesity-related proteins, in addition to total PCS and IS. Pearson’s correlation analysis showed that CI and ABSI were positively correlated with GDF15, FABP1, FGF21, hs-CRP, leptin, adiponectin, total PCS, and IS levels (Supplementary Table 2).

Effect of total PCS and IS on central obesity indices (CI and ABSI status) in the male patients: an SEM approach

Similar to the Pearson’s correlation coefficient analysis described above (Supplementary Table 2), we constructed an SEM to evaluate the impacts of uremic toxins (total PCS and IS) on central obesity indices (CI and ABSI status) (Fig. 3). The results showed that total PCS and IS had significant positive direct effects on central obesity indices (CI and ABSI), and positive direct effects on GDF15, FABP1, FGF21, hs-CRP, leptin, and adiponectin. Furthermore, total PCS and IS were found to have indirect influences on central adiposity metrics, such as WHR, mediated through hs-CRP. The estimated model demonstrated a good fit, as evidenced by a comparative fit index of 0.977, root mean square error of approximation of 0.042, and standardized root mean square residual of 0.038 (Fig. 3).

Fig. 3: Structural equation model for assessing central obesity in male patients with stable coronary artery disease.
figure 3

Comparative fit index (CFI), 0.977; goodness of fit index (GFI), 0.966; root mean square error of approximation (RMSEA), 0.042; standardized root mean square residual (SRMSR), 0.038. *p < 0.05, **p < 0.01, and ***p < 0.001. Path loadings are standardized coefficients. GDF growth differentiation factor, FABP fatty acid-binding protein, FGF fibroblast growth factor, Hs-CRP high sensitivity C-reactive protein, ADPN adiponectin, PCS p-cresylsulfate, IS indoxyl sulfate, CI conicity index, ABSI a body shape index, WHR waist-to-hip ratio.

Full size image

Discussion

The current study explored associations between circulating total PCS and IS concentrations with central obesity in patients with stable CAD. There were three key findings: (1) a significant positive correlation between total PCS and IS with WHR, CI, and ABSI in the male patients, whereas such correlations were not observed in the female patients; (2) an association between higher serum concentrations of total PCS and IS with CI ≥ 1.303 and ABSI ≥ 0.084 in the fully adjusted model in the male patients; and (3) SEM analysis revealed that uremic toxins, specifically total PCS and IS, had significant positive direct effects on central obesity indices (CI and ABSI), as well as on the six studied obesity-related proteins (GDF15, FABP1, FGF21, hs-CRP, leptin, and adiponectin). Furthermore, total PCS and IS were shown to have indirect influences on central adiposity metrics, including WHR, mediated through hs-CRP in the male patients.

The first key finding of this study is the association between total PCS and IS with WHR, CI, and ABSI in the male patients (Table 1). Both total PCS and IS originate from the intestinal microbial metabolism of dietary amino acids. PCS is derived from tyrosine, while IS is derived from tryptophan. After chemical modifications occur in the liver, both active metabolites enter the circulation and affect target organs [45, 46]. Previous studies have demonstrated that both PCS and IS directly contribute to vascular calcification in patients with CKD by activating inflammation and coagulation pathways [47]. In addition, they have been shown to be strongly linked with cardiovascular diseases and impaired glucose homeostasis [27, 47]. Furthermore, there is evidence that the accumulation of PCS is associated with the development of altered metabolic conditions such as ectopic lipid dysfunction and insulin resistance in the liver and muscles [29]. Moreover, the accumulation of IS has also been associated with changes in endothelial dysfunction, smooth muscle cell hyperplasia, thyroid function, vascular calcification, and an increase in atherosclerosis in men [48, 49]. However, the biological mechanisms through which total PCS and IS contribute to the pathogenesis of central obesity have yet to be elucidated. In this study, total PCS and IS were associated with central obesity in the male patients but not in the female patients. This result may be because the male patients had a significantly lower eGFR than the female patients (54.7 ± 20.2 ml/min/1.73 m2 vs. 63.4 ± 27.2 ml/min/1.73 m2, p = 0.0004; Supplementary Table 1). The accumulation of uremic toxins in the body is caused by a decrease in renal function [50], and previous reports have suggested their involvement in the regulation of adipose tissue [51]. Furthermore, we found that the male patients had a significantly higher WHR than the female patients (0.95 ± 0.07 vs. 0.89 ± 0.07, p < 0.0001; Supplementary Table 1). Further studies are needed to clarify whether sex differences are a factor in the association between total PCS and IS with central obesity.

The second finding of this study is the association between increased concentrations of serum total PCS and IS with CI ≥ 1.303 and ABSI ≥ 0.084 in the fully adjusted model in the male patients (Tables 4 and 5). Central obesity is known to cause disturbances in adipocyte biology and inflammation of adipose tissue, resulting in direct systemic metabolic consequences including insulin resistance, dysglycemia, alterations in blood pressure regulation, and lipid metabolism. These changes mutually favor endothelial dysfunction and atherogenesis [52]. Furthermore, chronic low-grade adipose tissue inflammation has been suggested to play a role in the progression of central obesity [53]. Moreover, an increase in systemic oxidative stress may play a significant role in the link between obesity, and central obesity in particular, with atherosclerotic cardiometabolic diseases. This may then indirectly or directly contribute to the development of atherosclerosis [54]. Koppe et al. proposed that PCS contributes to insulin resistance [29], and in vitro studies have indicated that the pathogenic mechanisms of PCS and IS originate from the promotion of reactive oxygen species. This activation triggers the nuclear factor-kappaB pathway, leading to both oxidative stress and the production of pro-inflammatory cytokines [55, 56]. In addition, IS is distributed in adipose tissue, where it can induce the expression of monocyte chemoattractant protein-1 (MCP-1) through the organic anion transporter/NADPH oxidase/reactive oxygen species pathway. This induction of MCP-1 by IS may facilitate macrophage infiltration into adipocytes, leading to an elevation in adipose tissue inflammation. Taken together, these findings suggest that IS primes adipocytes for an inflammatory response [57]. Our results support the concept [29, 55,56,57] that total PCS and IS may play a significant role in the pathophysiology of central obesity in male patients with stable CAD through insulin resistance, reactive oxygen species, and inflammatory responses.

The third key finding of this study is that SEM analysis revealed significant positive direct effects of uremic toxins (total PCS and IS) on central obesity indices (CI and ABSI), as well as on the six studied obesity-related proteins (GDF15, FABP1, FGF21, hs-CRP, leptin, and adiponectin) in the male patients. Nevertheless, total PCS and IS were found to have indirect influences on central adiposity metrics, including WHR, solely mediated through hs-CRP. Previous studies have shown that uremic toxins can contribute to a pro-inflammatory state [46], endothelial dysfunction [58], immune system activation [59], and oxidative stress [60]. These factors may lead to increased production of inflammatory markers, including hs-CRP. Megawati et al. also suggested that hs-CRP is related to overweight, obesity, and central obesity [61]. Taken together, these findings suggest an association between total PCS and IS with hs-CRP, thereby contributing to central obesity in male patients with stable CAD.

To the best of our knowledge, this is the first study to explore associations between total PCS and IS with central obesity in patients with stable CAD, including both males and females. The primary focus was on the male patients due to the observed associations between total PCS and IS with CI and ABSI in this group, whereas no such associations were found in the female patients. However, the inclusion of female patients in the analysis is also important. The findings contribute valuable insights into sex-specific associations, even though the smaller sample size of females may have limited the ability to detect significant associations in this group. Future research with a larger number of females are necessary to build on these initial findings and provide a more comprehensive understanding of these associations in women. The limitations to this study include the cross-sectional design, which restricts our ability to elucidate causal relationships between higher concentrations of total PCS and IS with central obesity. Hence, further studies with long-term follow-up are warranted to clarify the association between total PCS and IS with central obesity. Another limitation is that we exclusively enrolled patients with stable CAD, and thus our findings may not be readily generalizable to other groups. Further investigations incorporating larger and more ethnically diverse participants are necessary to verify our findings. Furthermore, the prognostic significance of elevated total PCS and IS concentrations in male patients with stable CAD and central obesity remains unknown, and further investigations are needed to clarify this issue. Moreover, CI and ABSI are indirect measurements and calculated indices of central obesity, for which they are not considered gold standard measurements. Finally, whether total PCS and IS are associated with other obesity-related proteins beyond hs-CRP in their contribution to central obesity among male patients diagnosed with stable CAD is unknown, and further studies are needed to elucidate this matter.

In conclusion, our study demonstrated an association between elevated circulating concentrations of total PCS and IS and central obesity, as assessed by ABSI and CI indices. Total PCS and IS appeared to play a role in the pathogenesis of central obesity through the mediation of hs-CRP among ethnically Chinese male patients diagnosed with stable CAD. These findings have important clinical implications. First, understanding the role of PCS and IS in central obesity can aid in developing targeted therapeutic strategies for male patients with CAD. By identifying and potentially modifying levels of these metabolites, healthcare providers may be able to reduce central obesity and its associated risks, leading to improved cardiovascular outcomes. Second, the partial mediation by hs-CRP highlights the significance of inflammation in the relationship between these metabolites and central obesity. This suggests that anti-inflammatory treatments might be effective in addressing central obesity in this patient population. Monitoring and managing hs-CRP levels could become an integral part of treating central obesity and preventing its complications in patients with CAD. Lastly, recognizing the sex-specific differences in these associations underscores the need for personalized medicine approaches. For female patients, in whom the associations were not significant, it may be necessary to explore other contributing factors or pathways to effectively manage central obesity. Future research with a larger number of females could provide more insights into these sex-specific mechanisms.