Relationship between visceral adipose tissue and triglyceride glucose index with liver status in men and women with metabolic syndrome features: sex differences

study association and and with A cross-sectional study was performed including 326 individuals with MetS (54–75 years) from the PREDIMED-Plus study. Liver markers, visceral fat (VAT) and the triglyceride glucose index (TyG) were assessed. Participants were stratied according to VAT and TyG tertiles. Receiver operating characteristic (ROC) curve was used to analyze the eciency of TyG for VAT.


Introduction
The metabolic syndrome (MetS) encompasses a cluster of cardiometabolic features like impaired glucose metabolism, dyslipidaemia, abdominal obesity, and elevated blood pressure [1]. The strong association between MetS and an increased risk of cardiovascular disease (CVD) as well as all-cause mortality is well documented [2,3]. Ageing and sex differences should be considered as well since it is a major contributor to the prevalence of cardiovascular and metabolic risk factors that constitute the syndrome [1,4]. On the other hand, non-alcoholic fatty liver disease (NAFLD) is recognized as the hepatic manifestation of MetS [5]. NAFLD is a highly prevalent chronic liver illness, whose incidence linearly increases with body mass index (BMI) and adiposity [6]. This condition is quite common in obese individuals with central adiposity [7,8]. The distribution of adipose tissue is of great importance since abdominal obesity is a key factor in the development of the MetS [9] and NAFLD [10] when differences between men and women are usually reported [11,12]. Insulin resistance is considered the primary triggering mechanism for the development of diabetes, NAFLD and MetS when fat accumulates in intraabdominal depots [9,10] Thus, body fat distribution in older adults and the differences between men and women is critical for determining how susceptible they are or will be to developing NAFLD and/or other CVD [11,[13][14][15] being partly attributed to sex differences in fat content [11,15]. In ageing, adipose tissues exhibited changes in quantity, distribution, hormone production, in ammatory status, and others. [11]. In this context, consistent evidence suggests that sex hormones play a major role in the activity and fat distribution, which has differential harm effects on metabolic status in male and female [11,13].
Central obesity is often quanti ed using waist circumference. But, it can be confounded by varying levels of subcutaneous fat in the waist, and may not accurately re ect visceral fat in all individuals [16]. The dual-energy X-ray absorptiometry (DXA) with CoreScan is a practical and valuable tool to assess visceral fat mass [17]. Nevertheless, the DXA equipment is expensive and might not be easy to access. In this sense, the identi cation of non-invasive markers able to discriminate subjects with higher visceral adiposity and higher susceptibility to develop NAFLD would be of great interest. Indeed, liver biopsy is the gold standard for the NAFLD diagnosis [7], but it is an invasive technique not suitable for routine screening and monitoring [7]. Several non-invasive markers related to liver status and insulin resistance have been proposed to characterize NAFLD [7,[18][19][20]. A novel potential marker is the triglyceride glucose index (TyG), which has demonstrated a better predictive value compared to fasting plasma glucose (FPG) for the risk of type 2 diabetes in normoglycemic individuals as well as to be related with insulin resistance [21]. In the present study, we hypothesized that the TyG index could be a good determinant in men and women with higher VAT and thus higher susceptibility to develop NAFLD. Therefore, the objective of the present work was to analyze the potentially detrimental effect of VAT accumulation on metabolic status and to assess the potential association of the TyG index with VAT, metabolic risk factors, NAFLD scores and serum markers in overweight/obese men and women with MetS and the differences between both sexes.

Study population and design
This research is a cross-sectional study concerning baseline data from al participant of the Navarra-Nutrition centre within the PREDIMED-Plus trial (ISRCTN89898870) (http://www.isrctn.com/ISRCTN89898870). PREDIMED-Plus is a multicenter, parallel-group, randomized trial carried out in Spain, aiming to evaluate the effectiveness of an energy-restricted traditional Mediterranean diet, physical activity promotion, and behavioural support (intervention group) on the primary prevention of CVD, in comparison with general advised energy-unrestricted Mediterranean diet (control group). Detailed methods and protocols of the study have been published previously [22,23]. In brief, 6874 individuals were recruited in 23 Spanish centres. Eligible participants were men (55-75 years) and women (60-75 years) with a BMI ≥27 and <40 kg/m 2 and ful lling at least three criteria for the MetS: waist circumference (WC) in Caucasian people ≥102 cm for men and ≥ 88 cm for women, elevated triglycerides levels ≥150 mg/dL or drug treatment for hyperlipidemia; reduced HDL-c <40 mg/dL in men and <50 mg/dL in women or drug treatment; elevated blood pressure systolic ≥ 130 and/or diastolic ≥85 mmHg or current use of antihypertensive medication; elevated fasting glucose ≥100 mg/dL or drug treatment, according to guidelines from the International Diabetes Federation/National Heart, Lung and Blood Institute/American Heart Association (2009) [24]. As described elsewhere, exclusion criteria included a background of alcohol overuse, liver injury, history of previous CVD, gastrointestinal or other disorders, infectious processes, therapy with immunosuppressive drugs, cytotoxic agents or systemic corticosteroids. The protocol and procedures were approved for the Research Ethic Committee for clinical investigations of the University of Navarra according to the Declaration of Helsinki. All participants provided written informed consent. At Navarra-Nutrition centre, a total of 422 participants were registered in the pre-inclusion period and 331 were included in the study, after excluding individuals, who did not meet inclusion criteria (n=2) and participants who refused to participate (n=89). For this study, we also excluded participants who did not have data for measured noninvasive biomarkers (n=5) and patients without DXA values (n=77) (Figure 1).

Study assessment
Clinical and biochemical measurements At baseline, participants completed an administered survey, which included questions about sociodemographic characteristics, lifestyle behaviours, diseases history, and medication. Smoking habit was classi ed into never, former, or current smoker as described elsewhere [22]. Blood pressure was measured in triplicate using a validated semiautomatic oscillometer (Omron HEM-705CP). Diabetes was established as previous diagnosis of diabetes or glycated haemoglobin (HbA1c) ≥6.5%, use of antidiabetic medication or fasting glucose ≥126 mg/dl according to the American Diabetes Association (ADA) guidelines [25]. After overnight fasting for at least 12-h, a blood sample was obtained from each participant. Serum and plasma were collected and frozen at -80˚C. All biochemical measurements, including plasma glucose, HbA1c, insulin, total cholesterol, high-density lipoprotein cholesterol (HDL-c), triglyceride, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and gamma-glutamyl transferase (GGT) were performed using standard laboratory enzymatic methods and following validated protocols [22]. The broblast growth factor 21 (FGF-21) plasma concentrations were measured using human FGF-21 Quantikine ELISA Kit (R&D Systems, Minneapolis, MN, USA) with an autoanalyzer system (Triturus, Grifols SA, Barcelona, Spain) following the manufacturer's instructions. Low-density lipoprotein cholesterol (LDL-c) concentration was calculated by Friedewald's formula and the very-low-density lipoprotein cholesterol (VLDL-c) were calculated as triglycerides/5 [26]. Also, homeostatic model assessment for insulin (HOMA-IR) was calculated according to the formula: fasting insulin (mIU/L) x fasting glucose (nmol/L)/22.5 [27].

Dietary variables
Trained dietitians face-to-face administered a semi-quantitative 143-item food frequency questionnaire to estimate energy intake and alcohol consumption [28]. Also, a 17-item questionnaire, which is a modi ed version of the previously validated questionnaire used in the PREDIMED study to assess the participant´s adherence to the Mediterranean diet was implemented [29].

Physical activity measurement
Physical activity was assessed using the short REGICOR (Registre Gironi del Cor) questionnaire that showed high reliability and sensitivity to detect changes in moderate and vigorous intensity [30,31]. This tool was validated in the Spanish adult population, which is a version of the Minnesota Leisure Time [30]. This questionnaire evaluated the total energy expenditure in leisure-time physical activity using Metabolic Equivalent Tasks (METs) in minutes/week. Physical activities were classi ed into lightintensity (<4 METs), moderate-intensity (4-5.5 METs), and vigorous-intensity (≥6 METs) as detailed in the report [30]. Sedentary lifestyles were evaluated considering a validated Nurses' Health Study questionnaire [32]. For the present study, physical activity was expressed as MET/hours/week.

Anthropometry and body composition measurements
Anthropometric measurements were performed by trained dietitians following standardized PREDIMED-Plus protocols [22]. Weight, height and WC, were measured using a calibrated scale, a stadiometer and an anthropometric tape, respectively. BMI was conventionally calculated as weight in kilograms divided by the height in square meters (kg/m 2 ). VAT was estimated using dual-energy X-ray absorptiometry (Lunar iDXA ™, software version 6.0, Madison, WI, USA) connected with enCore ™ software, which was assessed by trained operators according to standard procedures supplied by the manufacturer.

Non-invasive markers
The TyG index is a newly described marker that has been reported as a useful screening tool for insulin resistance [21,33,34], NAFLD [35] and as an early predictor of MetS features [36]. This marker was calculated using biochemical data according to the following formula [33], The HSI was validated in a cohort of patients with NAFLD diagnosed by ultrasonography [37]. The HSI equation: [37,38] it was also computed to estimate liver status. Another liver marker as an indicator of NAFLD is the FLI, which was calculated as as previously described [39].

Statistical analyses
Continuous variables are presented as means ± SD and categorical variables as numbers (n) and percentages (%). One-way analysis of variance (ANOVA) and chi-square tests were used to assess differences between groups, as appropriate. The ANCOVA test after adjustment for the following potential confounders: age (years), physical activity (MET/min/week), energy intake (kcal/d), alcohol intake (g) and smoking status (never, former, current). Bonferroni correction was applied to assess differences in metabolic and liver parameters according to sex-speci c VAT tertiles. Crude and multiple linear regression models adjusted by age (years), sex (male and female), physical activity (MET/min/week), energy intake (kcal/d), alcohol intake (g) and smoking status (never, former, current) were tted to analyze the association between NAFLD biomarkers and the TyG index. Tests of linear trend were assessed assigning the median value of each tertile of TyG and then using it as a continuous variable and correlation was assessed using the Pearson's coe cient. The area under the ROC curve (AUC) was performed to quantify the diagnostic accuracy of TyG index as a predictor of VAT considering as references values of 50 th percentile of VAT by sex. All tests were two-sided and the cut-off level of signi cance was de ned at 0.05. Statistical analyses were carried out with Stata 12.0 software (StataCorp LP, College Station, TX).

Study sample characteristics
Baseline characteristics of men and women according to VAT sex-speci c tertiles are summarized in Table 1. As expected, BMI and WC increased across VAT tertiles in men and women. No signi cant differences were found in the frequency of diabetes, hypertension and smoking habits among tertiles in both sexes. Likewise, blood pressure (SBP and DBP) measurements, energy intake, alcohol consumption, adherence to the Mediterranean diet score, and physical activity did not differ statistically.

Crosstalk between VAT, TyG index and NAFLD risk factors in both sexes
Anthropometric, metabolic pro le and liver status of participants are reported in Table 2. The adjusted analysis revealed that BMI and WC were signi cantly increasing through VAT tertiles speci c by sex. Moreover, insulin, TyG and HOMA-IR increased with VAT tertiles reaching statistical differences among them (all p<0.001). Glucose and HbA1c did not show differences between tertiles. As concerns lipidic markers, the T3 group presented signi cantly higher levels of VLDL-c [mean 32 3)] than T1 participants, while no associations were found regarding total cholesterol, LDL-c and HDL-c serum levels. Men and women in the highest VAT tertile showed signi cantly higher ALT levels, HSI and FLI scores as compared with participants in the lowest tertile of VAT. No signi cant differences were found in AST and FGF-21 levels in VAT tertiles.
The association of the TyG index with variables related to liver health was explored (Table 3). Linear regression models were tted considering NAFLD related markers as dependent factors and the TyG index as the independent variable (  Receiver operator characteristic (ROC) analyses for the TyG index to predict VAT in male and female ROC curves were applied to assess the capacity of the TyG index to identify elevated VAT accumulation in both sexes (Figure 3). The AUCs of the TyG index for prediction of VAT were 0.570 (95% CI: 0.48 to 0.66) for men and 0.713 (95% CI: 0.62 to 0.79) for women, which appears as a relevant outcome of their analyses.

Discussion
In this cross-sectional study, VAT and the TyG index were associated with relevant metabolic and liver risk factors linked to NAFLD in male and female with MetS. Moreover, the TyG index could be a reliable indicator of visceral adipose dysfunction in women (AUC: 0.570), but not in men (AUC: 0.713). Many metabolic abnormalities related to insulin resistance are often occurring in obese individuals with higher amount of VAT [9,40]. In our study, men and women with increased VAT showed higher levels of serum variables related to NAFLD. Interestingly, the TyG index and atherogenic lipid pro le (VLDL-c, triglycerides and TG/HDL-c ratio) were signi cantly increased across tertiles of VAT speci c by sex independently of confounder factors. In line with our results, Lee and colleagues observed that VAT and triglycerides were independent risk factors for hepatic steatosis [41]. VAT as the main source of free fatty acids (FFAs) and other biological compounds, which entering portal circulation into the liver contribute to hepatic fat accumulation [40]. Younger women have the ability to partition FFA's towards ketone body production rather than VLDL-triacylglycerol, but is affected in postmenopausal women that have a negative impact on liver status [42]. Moreover, a statically signi cant increase of liver markers (ALT, FLI and HSI) in both sexes with excessive accumulation of VAT depots was found. In agreement with our results, Chung et al. indicated that increased VAT was associated with higher ALT levels [43] and NASH or signi cant brosis in subjects diagnosed with NAFLD [44]. Based on these data, central adiposity plays a key role in NAFLD pathogenesis [10,15,45] promoting liver damage [44], insulin resistance and disrupted lipid metabolism [46].
Nowadays, NAFLD has become a public health problem with a negative impact over the individual's health, socioeconomic and health care system [6,47]. In this context, early screening is crucial in the NAFLD pathogenesis [7]. Liver biopsy is the gold standard for NAFLD diagnosis [7]. However, it has several limitations such as sampling error, cost, medical complications, and technical di culties [48]. In this regard, many techniques have been used in the detection and featuring NAFLD that showed to be relatively effective, inexpensive and useful in a primary health care setting [20,[48][49][50]. The TyG index is a novel marker that has exhibited a good accuracy for recognizing insulin resistance [21,51]. Furthermore, this marker showed highly sensitive for detecting NAFLD [35]. Interestingly, the multivariable regression analysis evidenced that individuals with higher TyG (> 9.1) value were associated with higher levels of HOMA-IR ALT, GGT, FGF-21, and HSI units compared with lower TyG values (≤ 8.7 units), after adjusting for sex and potential confounders. These results could be partially explaining by that NAFLD is prevalent in men but this trend increases in postmenopausal women, which is related to regional adiposity and hormones effects in ageing [15,42]. Serum aminotransferase levels might associated with insulin resistance [52]. Bonnet et al. found that increased levels of ALT and GGT are strongly associated with hepatic insulin resistance and decreased hepatic insulin clearance [53]. The FGF-21 is primarily produced in hepatocytes and is implicated in the regulation of glucose-lipid metabolism, insulin sensitivity, in ammation and energy homeostasis [54]. Several clinical studies and reviews have documented that disrupted adipose tissue and excessive intrahepatic fat accumulation may trigger FGF-21 resistance [54, 55]. Thus, Shen and colleagues [56] found that NAFLD patients showed signi cantly higher serum FGF-21 levels compared with subjects without NAFLD [56]. The FLI was positively associated with the third tertile of the TyG index, in the same manner that the HSI, indicating that subjects with higher values of TyG had 3.46 more units of HSI compared to the reference (lower values). Taking together, these results can be explained by the fact that insulin resistance is a major feature of NAFLD by increasing de novo lipogenesis and FFAs ux to the liver through decreased inhibition of lipolysis [5] promoting in ammation, oxidative stress [57] and hepatocyte injure [58].
DXA has been considered the gold standard for body composition measurements [17]. Nevertheless, this imaging technic for assessing adipose tissue distribution is expensive and not feasible for community screening. In our results, we observed a close relationship between insulin resistance and dysfunctional VAT. Interestingly, men had higher amounts of VAT than women. Meanwhile, women and men with ≥ VAT median had similar TyG values (data not shown). Moreover, the ROC curves indicate a moderate predictive ability of the TyG to discriminate VAT in women (AUC = 0.713), but it was weak for men (AUC = 0.570).
The connection between body fat distribution and adipose tissue biology with insulin resistance differ by sexes, age and other factors [13]. In general, women have more total body fat mass and men presented higher abdominal/visceral fat mass [13]. However, decreased levels of estrogen and adipose tissue re-distribution by increased depots of VAT are characterized in postmenopausal women [11,13]. Estrogen and testosterone are involved in glucose and lipid metabolism. A disbalance of sex hormone levels promote insulin resistance and an atherogenic lipid pro le, which increased the risk of CVD in older women [13]. Interestingly, some studies have suggested that obese women are more insulin sensitivity than men despite a higher amount of VAT [11]. However, the plausible mechanism is still unclear. Recently, the Netherlands Epidemiology of Obesity Study (NEO) showed that in obese women, VAT was differently associated with cardiometabolic risk factors as compared obese men [12]. But, this is in contrast with Ferrara et al. who reported that older obese men are more insulin resistant compared with older women even adjusted for differences in abdominal fat distribution measured by DXA [14]. This result can be attributed to the limited number of participants. One possible explanation for our results could be that women exhibit a greater amount of FFAs delivery derived from VAT lipolysis [59]. Thus, there are differences between sexes in the lipid storage capacity and function [57]. Serra et al. showed that postmenopausal women (overweight or obese) diagnosed with MetS had lower adipose tissue LPL activity and limited capacity for lipid accumulation in subcutaneous abdominal adipose tissue leading to higher rates of lipid, accumulation of VAT and insulin resistance [60]. These results reinforce that a high amount of VAT has a negative impact over metabolic and liver status leading to elevated insulin levels and hepatic insulin resistance [5,45]. In this context, the improvement of the knowledge of these interrelationships in male and female with MetS should be useful to easily identify individuals with a high risk of NAFLD, which may allow early intervention and prevention of NAFLD complications.
The strengths of this study include the relatively large sample size of individuals with MetS in the framework of PREDIMED-Plus. Also, VAT was objectively measured with a validated imaging technique. However, some limitations required considerations. First, the cross-sectional design cannot imply a causal relationship. Second, the lack of NAFLD diagnosis by liver biopsy or imaging techniques, but it is important to note that liver biopsy is not available or feasible in large epidemiological studies. On the other hand, validated non-invasive markers were used to estimate hepatic fat accumulation [38].

Conclusion
A higher accumulation of VAT is associated with cardiometabolic and liver risk factors linked to NAFLD in older men and women with overweight/obese and MetS. Moreover, we reinforce that in addition to anthropometric measurements such as WC or DXA approach, the TyG index could be a useful simple marker to identify dysfunctional VAT phenotype in women with MetS better than in men. Research Ethics Committees from all recruitment centers approved the study protocol following the rules of the Declaration of Helsinki on July 24, 2014 (approval number ISRCTN898988709). All participants were informed of procedures and signed an informed consent.

Consent for publication
Not applicable.

Page 22/24
Abbreviations: WC, waist circumference; HOMA-IR, homeostatic model assessment for insulin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; FGF-21, fibroblast growth factor-21; HSI, hepatic steatosis index; FLI, fatty liver index. Figure 1 Study ow for the selection of the study population Correlations between VAT and parameters related to glucose and insulin homeostasis in men and women diagnosed with MetS