Estimated visceral adiposity is associated with risk of cardiometabolic conditions in a population based study

Visceral adiposity is a major risk factor of cardiometabolic diseases. Visceral adipose tissue (VAT) is usually measured with expensive imaging techniques which present financial and practical challenges to population-based studies. We assessed whether cardiometabolic conditions were associated with VAT by using a new and easily measurable anthropometric index previously published and validated. Data (1529 participants) came from the European Health Examination Survey in Luxembourg (2013–2015). Logistic regressions were used to study associations between VAT and cardiometabolic conditions. We observed an increased risk of all conditions associated with VAT. The total adjusted odds ratio (AOR, [95% CI]) for hypertension, prediabetes/diabetes, hypercholesterolemia, and hypertriglyceridemia for the fourth quartile of VAT compared to the lowest were 10.22 [6.75, 15.47]), (5.90 [4.02, 8.67]), (3.60 [2.47, 5.25]) and (7.67 [5.04, 11.67]. We observed higher odds in women than in men for all outcomes with the exception of hypertension. Future studies should investigate the impact of VAT changes on cardiometabolic health and the use of anthropometrically predicted VAT as an accurate outcome when no biomedical imaging is available.

Differences were observed between men and women. Men were more likely to have higher BMI, WC, and anthropometrically predicted VAT while women had higher thigh circumference. Men were more likely to have higher systolic and diastolic blood pressure, LDL cholesterol, triglycerides and fasting plasma glucose levels. Men had almost twice as much hypertension (40.09%), combined prediabetes and diabetes (42.63%) and hypertriglyceridemia (40.17%) as in women (23.40%, 22.16% and 19.09% respectively). Women smoked and consumed alcoholic drinks less than men.
Cardiometabolic conditions by quartiles of VAT. The proportion of hypertension, combined prediabetes and diabetes, hypertriglyceridemia and metabolic syndrome increased with VAT quartiles in both men and women ( Table 1). The largest prevalence gradient was observed for metabolic syndrome in both men (from 1.7% in the first quartile to 82.1% observed in the fourth quartile) and women (from 1.1% in the first quartile to 62.2% observed in the fourth quartile). The proportion of hypercholesterolemia increased with VAT in both men and women. We observed that median values of WC and thigh circumference increased with VAT, but when we visualized VAT for similar WC values (Supplementary Figure S1), we observed that thigh circumference decreased for both men and women. Similar results were observed in the prevalence of cardiometabolic conditions by quartiles of WC (Supplementary Table S2).
VAT association with all cardiometabolic conditions. Results from logistic regression analysis examining the association between anthropometrically predicted VAT and cardiometabolic conditions are presented in Table 2. We observed an increase in the odds of all metabolic and cardiovascular conditions associated with VAT in both men and women. The strength of the association was reduced but remained statistically significant after adjusting for socioeconomic status (education and employment status) (model 1), and lifestyle (model 2). The association observed was strongest in men for hypertension: adjusted OR [95% CI] for men were 2.51 [1.46, 4.29], 4.08 [2.40, 6.93], and 11.83 [6.82, 20.49] for the second, third and fourth quartile of VAT. For women, the values were 1.93 [0.92, 4.00], 3.41 [1.69, 6.85], and 8.21 [4.12, 16.36] for the second, third and fourth quartile of VAT. Nevertheless, women observed a strongest association for combined prediabetes and diabetes 7.57 [3.93, 14.59] for the fourth quartile of VAT in women compared to 5.41 [3.26, 8.97 in men), hypercholesterolemia (5.28 [3.09, 9.00] for the fourth quartile of VAT in women compared to 2.26 [1.33, 3.84] in men) and hypertriglyceridemia (14.62 [6.30, 33.90] for the fourth quartile of VAT in women compared to 6.78 [3.97, 11.56] in men.

Discussion
The present study highlighted an increase of all metabolic and cardiovascular conditions associated with anthropometrically predicted VAT in adults aged 25-64. The association observed was independent of socioeconomic status and lifestyles. Our findings confirm that VAT is a major independent predictor risk factor of cardiometabolic risk as observed in previous epidemiological studies 22 . This can be explained by the high metabolic activity of VAT and its pro-inflammatory activity (production of cytokines with inflammatory effects and blocking of those anti-inflammatory) 23,24 . Moreover, compared to other fat deposits, VAT has larger and dysfunctional adipocytes, which are less insulin sensitive and with increased lipolytic activity. As the adipocytes grow, they accumulate triglycerides, becoming leptin resistant and promoting the synthesis and release of free fatty acids 23 . We observed that both WC and thigh circumference increased with VAT, but as previously reported, when WC and age were constant thigh circumference decreased with VAT for both men and women 25 . This is in line with previous evidence showing that VAT is the major risk factor of cardiometabolic morbidity and premature mortality, while lower-body fat mas plays a protective role and should be maintained when reducing VAT 26 . We observed sex differences in cardiometabolic conditions with men having a higher prevalence of all conditions compared to women, in line with previous evidence among middle-aged adults 27       www.nature.com/scientificreports/ diabetes and prediabetes and hypertriglyceridemia prevalence were almost twice as high in men compared to women. Metabolic syndrome was 1.5 times higher in men compared to women. Closely related to these results are differences observed in WC and VAT being both higher in men compared to women as well as certain risk behaviours and socioeconomic differences such as lower consumption of alcohol and cigarettes and lower socioeconomic status in women compared to men. Sex differences on VAT are expected, since men are characterized by having a greater concentration of fat in the abdominal area compared to women that usually concentrates in the thighs and hip (gluteo-femoral pattern) 29 . Luxembourg is a small European country whose cultural diversity-nearly one in two residents is of foreign origin-accounts for the country's well-documented heterogeneous and complex health profile. As in other high-income countries, in Luxembourg CVD is the leading cause of death (31.8%) while deaths due to endocrine, nutritional and metabolic diseases have more than doubled in the past 15 years, in both men and women 30,31 . Results from the present study show that compared to a previous study conducted in 2007 in Luxembourg, no reduction in cardiometabolic conditions has been observed over the last decade 32 and even an increase has been noted in certain conditions such as diabetes or metabolic syndrome 33 . This could explain why cardiovascular diseases remained the main cause of mortality in Luxembourg in 2016 30 .
These results provide compelling evidence on the current burden of cardiovascular and metabolic conditions in Luxembourg in both men and women, and the need for public health initiatives to alleviate the societal impact of these highly prevalent disease conditions. Moreover, VAT management should be considered as a privilege area of study to tackle metabolic and cardiovascular health issues. As reported in other studies 34,35 , we observed that cardiometabolic conditions were more prevalent among individuals with poor nutritional status, smoking, consuming alcohol, and with sedentary habits. As observed by Shi et al. 2011, abstention from smoking, regular physical activity, and a moderate consumption of alcohol were related with less cardiometabolic conditions 34 . Moreover, studies have observed that a combination of a healthy diet and physical activity (regardless of the quantity/intensity) have a strong effect on reducing VAT, even with minimal weight loss, thus representing a cost effective non-pharmacological intervention to reduce the impact of VAT on cardiometabolic health 22 . At present, there is no specific treatment to reduce VAT without also reducing lower-body fat mass. There are only experimental clinical studies in progress with the aim of personalizing treatments to each specific situation/ individual 36,37 . Studies also observed an effect of socioeconomic conditions, with those with lower socioeconomic status being at higher risk of developing cardiometabolic diseases [38][39][40] . Both lifestyle and socioeconomic characteristics explained in part, but not completely, the association between VAT and cardiometabolic conditions, as we observed that the association remained statistically significant even after adjusting for those factors.
Although there is evidence showing that the socioeconomic effect could be mediated by health behaviours (e.g. smoking) 35 , we observed two independent effects (model 1 and model 2). Results of this study must be interpreted with caution, taking into account the following limitations. The design of the present study was cross-sectional, hence no temporal relationship or causality can be inferred. The participation rate was rather low yet still representative of the target population 41 . VAT was measured indirectly. Instead of using biomedical imaging techniques (e.g. MRI, CT-Scan), we estimated VAT with anthropometric measurements. Nevertheless, the predictive anthropometric models of VAT used in the present study were previously developed and validated as the most accurate predictor of biological cardiometabolic risk factors, all-cause and cause-specific mortality in Europid descendants, when biomedical imaging data are not available [17][18][19] . Results www.nature.com/scientificreports/ from these studies observed a high correlation of VAT (assessed by imaging techniques) with anthropometric VAT models, whereas other studies observed that WC was higher correlated with SAT and fat mass than with VAT 42 . Finally, we did not have information on other potential biomarkers of cardiometabolic risk such as markers of inflammation. In summary, anthropometrically predicted VAT was associated in the present work with all metabolic and cardiovascular conditions in both men and women even after adjusting for socio-demographic and behavioural characteristics. This reinforces the role played by VAT as a major independent risk factor for cardiometabolic health. Anthropometrically predicted VAT should be used in future epidemiological studies to investigate metabolic and cardiovascular disease when no biomedical imaging measurements are available and replicated in other contexts /populations. Likewise, prospective and intervention studies should place greater focus on the impact of changes in VAT on cardiometabolic health.  www.nature.com/scientificreports/ While objective measures of VAT (e.g. CT-Scan, MRI) are not covered by EHES-LUX, the survey does dispose of accurate and complete set of anthropometric measurements. In the present study we excluded 5 individuals with values of visceral adipose tissue inferior or equal to zero. One individual did not have a measure of height and thus VAT was not possible to calculate. The final sample size of the present study was 1441 participants. The study was approved by the National Research Ethics Committee (CNER, No. 201205/07) and notified to the Luxemburgish National Commission for Data Protection (CNPD). All methods were performed in accordance with the relevant guidelines and regulations.

Methods
Cardiovascular and metabolic conditions. Hypertension was defined as systolic/diastolic blood pressure of ≥ 140/90 mm Hg, self-report of a physician diagnosis, or on antihypertensive medication 41 . Fasting plasma glucose, haemoglobin A1c, total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and triglycerides were measured in participant's blood samples 41 . Prediabetes and diabetes were defined as fasting glucose of ≥ 100 mg/dL, self-report of physician diagnosis, or on antidiabetic medication. Hypercholesterolemia was defined as total cholesterol ≥ 190 mg/dL or on medication to reduce cholesterol. High LDL-C was defined as LDL-C ≥ 115 mg/dL for both men and women 43 . Low HDL-C was defined as HDL-C < 50 mg/dL for women and 40 mg/dL for men. In this paper, we use the term hypercholesterolemia in reference to total cholesterol. Hypertriglycemia was defined as triglycerides ≥ 150 mg/dL or on medication. Metabolic Syndrome was defined following the International Diabetes Federation 44 . This definition includes the presence of central obesity WC ≥ 94 cm for men and WC ≥ 80 cm for women) and at least two of the following factors: total triglycerides ≥ 150 mg/dL or on medication, low HDL-C (< 40 mg/dL for men, < 50 mg/ dL for women) or on medication, high blood pressure (systolic blood pressure ≥ 130 mm Hg or diastolic blood pressure ≥ 85 mm Hg) or on medication, high fasting plasma glucose (≥ 100 mg/dL) or self-report of a physician diagnosis.
Anthropometric variables. Weight  Covariates. Based on the literature review we selected a list of potential covariates 27,34,38 . Demographic characteristics included age and sex (men and women). Lifestyle characteristics included smoking status (current smoking or quit < 12 months vs non-smokers or quit > 12 months), alcohol consumption (non-alcohol consumption, ≤ 6 drinks/week, > 6 drinks/week) and aerobic physical activity (min/week). For both alcohol and physical activity, we used validated questionnaires with standardized questions for European populations from the European Health Interview Survey (EHIS). Socioeconomic characteristics included education (tertiary education vs secondary and primary), and job status (employed vs not employed). To analyse associations between cardiometabolic outcomes (e.g. hypertension, prediabetes and diabetes and total cholesterol) and covariates, we used a Pearson's chi-squared test (for probabilities related to frequencies) or Wilcoxon-Mann-Whitney U two-sample test (for probabilities related to medians) to compare characteristics between men and women. We used non parametric test because data was not normally distributed.

Statistical data analysis.
Distributions of cardiometabolic conditions across VAT quartiles were measured with the Cochran-Armitage P-trend test for categorical variables and Jonckheere-Terpstra test for continuous variables. We performed multivariable logistic regression analyses to study the association between VAT quartiles and cardiometabolic outcomes in unadjusted (Model 1) and adjusted models for education and employment status (Model 2), and lifestyle (e.g. smoking, alcohol consumption and physical activity) and socioeconomic conditions (Model 3). All analyses were stratified by sex, given the well-known differences in visceral adiposity distribution and cardiometabolic disease prevalence between women and men 46 . Only variables with a P-value < 0.20 in univariate analyses were included in the final model. Although the main objective of the paper was to analyze VAT quartiles www.nature.com/scientificreports/ related to cardiometabolic outcomes, we performed additional analyses dividing individuals by quartiles of waist circumference. The aim was to assess whether the results were similar, better, or worse than those obtained with estimated VAT (Supplementary Table S2).We used the Akaike information criterion (AIC) to evaluate the model fit quality of the univariate analyses using VAT and WC quartiles. We calculated WC score quartiles for both men and women as follows: Q1 (≤ 88.50), Q2 (88.51-95. 45 Table S3). The number of events per variable in the multivariable logistic regression were greater than 10 47 . Multicollinearity between covariates were tested. Weighted regression was used to correct for possible heteroscedasticity. A twotailed P-value < 0.05 was considered statistically significant. Analyses were performed using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

Data availability
The data generated during and/or analysed during the current study are available from the corresponding author on reasonable request.