Visceral adiposity indicators and cardiovascular risk in hemodialytic patients

ABSTRACT Objective: Cardiovascular diseases represent the main cause of death in chronic kidney disease (CKD). We aimed to evaluate the prevalence and association of the hypertriglyceridemia-waist phenotype (HWP) and visceral adiposity index (VAI) with cardiometabolic risk factors (CR) in patients with CKD on hemodialysis (HD). Materials and methods: The study is based on a cross-sectional design with 265 HD patients in two cities in northeastern Brazil. The VAI was calculated considering the variables body mass index (BMI), waist circumference (WC), triglycerides (TG) and high density lipoprotein cholesterol (HDL-c). HWP was defined as the concomitant elevation of WC and TG. The Poisson Regression Model with robust variance estimation was adjusted considering a hierarchical approach for explanatory variables. Prevalence ratios (PR) were also estimated. The level of significance adopted was 5%. Results: In our study HWP and VAI prevalence’s were 29.82% and 58.49%, respectively. In the final model, there was an association between VAI and female gender (PR = 1.46; p < 0.0001) and high body fat (% BF) (PR = 1.33; p < 0.0019). HWP was associated with females (PR = 1.80; p = 0.002), alcohol consumption (PR = 1.58; p = 0.033), obesity (PR = 1.89; p = 0.0001), high %BF (PR = 1.76; p = 0.012) and reduced HDL-c (PR = 1.48; p = 0.035). Conclusion: The HWP stood out as the association with more CR factors, representing a promising method for tracking cardiometabolic risk in HD patients, mainly female.


INTRODUCTION
C hronic kidney disease (CKD) is a public health problem defined as an abnormality of kidney structure or function present for more than three months, with health implications (1). Among patients with CKD, cardiovascular disease (CVD) is the main cause of death at any stage of the disease (2).
Although traditional cardiovascular risk factors (CR) cannot alone explain the high risk of CVD presented by patients on hemodialysis (HD), they seem to be highly predictive of this nosological entity (3). In this context, the investigation of the role of visceral adiposity has gained prominence since it has been associated with metabolic abnormalities in patients under dialysis treatment (4).
Predictive equations seek to find a practical and accessible marker based on indirect measurements of visceral adiposity, among which are visceral adiposity index (VAI) and the hypertriglyceridemia-waist phenotype (HWP) as identification tools of visceral adipose dysfunction and atherogenic metabolic triad (5)(6).
VAI has been reported as an indicator associated with long-term cardiovascular outcomes. All-cause mortality (7) and metabolic syndrome (8). HWP manifests as worse carotid atherosclerosis in patients with CKD (9).
Thus, taking into account the impacts of visceral adiposity on cardiovascular events frequently found in the HD scenario, in addition to the scarcity of studies assessing such indexes in the Brazilian HD population and, above all, in the northeastern region of Brazil, where resources are scarce, the present study aims to assess HWP prevalence and visceral adipose dysfunction according to VAI and to analyze its association with traditional CR factors in patients with CKD on HD.

Study design
This is a cross-sectional study that performs the analysis of two databases from two studies that evaluated patients with CKD undergoing dialysis at two capitals in the northeast of Brazil (São Luís, MA, and Recife PE) from January to December 2016 (10)(11).

Ethical aspects
The present study obtained approval from the Research Ethics Committee involving Human Beings of the Hospital das Clínicas of the Federal University of Pernambuco (CAAE 25657819.0.0000.8807) in accordance with the Resolution no. 466/2012. All study participants signed the informed consent.

Sample, inclusion and exclusion criteria
Clarifications regarding the population and inclusion criteria for the previous studies carried out in the cities of São Luís and Recife have been reported in previous studies (10)(11). In brief, patients were regularly registered in the dialysis program undergoing renal replacement therapy for at least three months and for least three-hour dialysis sessions of both genders, aged 18 years or over, who signed the Informed Consent.
The following category of patients were not included: pregnant, amputated, those with neurological diseases or stroke sequelae, autoimmune and infectious diseases, cancer, and acquired immunodeficiency syndrome. Volunteers were recruited during HD sessions held in 2016 at São Luís (MA) by stratified random sampling and at Recife (PE) by convenience. At that time, the objectives, risks, benefits, procedures adopted in the research, and study eligibility criteria were explained. Eligible patients obtained express authorization to participate in the study by signing the Informed Consent. All evaluated patients who had fulfilled the interest variables for the present study were simultaneously included.

Sample calculation
For those studies, the sample was calculated a posteriori using the software OpenEpi, version 3.01, considering the sum of HD patient populations of the seven evaluated centers, the prevalence reported in the literature of 24.6% (12), and adopting a 95% level of significance and a 80% test power. Thus, the minimum sample required would be 237 patients.
The sum of databases from both studies included data from 468 patients. Since only 265 patients fulfill the prerequisite of having all interest variables, these comprised the study sample.

Sociodemographic and lifestyle variables
The following sociodemographic characteristics were analyzed: age, gender, self-reported color (white, black, brown, and others), monthly family income (described in minimum wages -MW -for 2016), education (dichotomized in up to nine years of study and more than nine years of study), and marital status (dichotomized in having a partner or without a partner). This information was self-reported by the patients in an interview and/or collected from clinical records.
Smoking and alcohol consumption were considered as lifestyle variables, which were categorized as yes and no.

Clinical and biochemical variables
The duration of HD treatment (in <five months, ≥five months, and <12 months, and ≥12 months) (13) and the presence of associated comorbidities, such as systemic arterial hypertension (SAH) and diabetes mellitus (DM), self-reported in interviews or collected from clinical records (dichotomized as yes or no), were considered as clinical variables. The lipid profile was also evaluated, in which the following variables were considered: total cholesterol (TC), triglycerides (TG), low density lipoprotein cholesterol (LDL-c), and high density lipoprotein cholesterol (HDL-c). For the purpose of classifying the lipid profile, the criteria recommended by the National Kidney Foundation (NKF-KDOQI) (14) was considered.

Anthropometric and body composition variables
The body mass index (BMI) was obtained by the dry weight quotient (in kg, measured using a calibrated scale, Filizola ® , Brazil) and height (in square meters, measured by a stadiometer, Alturexata ® , Brazil), adopting the classification proposed by World Health Organization (WHO) for adults (15). For the elderly, the classification used was proposed by Lipschitz (16). Waist circumference (WC) was measured using an inextensible measuring tape (Sanny ® , Brazil) at the midpoint between the last rib and the iliac crest using the cut points recommended by the International Diabetes Federation (IDF) (17). All anthropometric measurements were performed after the HD session, as recommended by the National Kidney Foundation (NKF-KDOQI) (18).
Body fat percentage (%BF) was evaluated using an electrical bioimpedance (BIA) tetrapolar Biodynamics ® equipment 30 minutes after the dialysis session (18). Data such as weight, height, and level of physical activity were previously fed into the device and using the prediction equations of the software from the device itself, the BF was estimated. For %BF classification, %BF ≥ 25 was the threshold for men and ≥ 32 for women (19).

Visceral adiposity index (VAI)
The VAI was calculated according to the equation proposed by Amato and cols. (6) which is genderspecific, where WC is expressed in cm, BMI in kg/m², and TG and HDL-c in mmol. The latter were obtained by multiplying the TG and HDL-c in mg/dL by 0.0113 and 0.0259, respectively (20).
Men  (21). For analysis purposes, the classification was dichotomized into "increased VAI levels" or "normal" VAI levels, where "increased VAI levels" refers to VAI values classified as low, moderate, and high according to age.

Hypertriglyceridemia waist phenotype (HWP)
The HWP was defined according to the method proposed by Lemieux and cols. (5), which characterizes it from a concomitant increase in WC and TG. The cutoff points of the IDF (17) and the NKF-KDOQI (14) were adopted to classify WC and TG, respectively (≥80 cm for women and ≥90 cm for men, and TG > 150 mg/dL). The patients were grouped into "Presence" or "Absence" of HWP according to the mentioned criteria.

Statistical analysis
The characteristics of the patients evaluated were presented in frequencies and percentages, and the proportions were compared using the Chi Square test or Fisher's Exact test. A Poisson multivariate regression model with robust variance according to a hierarchical approach (22) was adjusted at four levels to assess the associated factors of HWP and VAI. The first level corresponded to sociodemographic variables (age, gender, and self-reported color). The second level considered the variables related to lifestyle (alcohol consumption and smoking) and the third level considered the anthropometric and body composition measurements (BMI and %BF). The fourth level was composed of clinical variables (DM, SAH, TC, HDL-c, and LDL-c).
The variables related to each level were inserted into the model simultaneously, and those with a p-value lower than 0.05 remained in the analysis at the other levels, even if the p-value was no longer significant. After the analysis of the fourth level, the variables that obtained a p-value lower than 0.05 remained in the final model. Prevalence ratios and their respective 95% confidence intervals were also estimated. The variables WC and TG did not enter the regression model for the VAI and HWP, nor did HDL-c and BMI in the regression model for the VAI, since these variables make up their respective predictive equations. In all analyses, a significance level of 5% was considered and the statistical program used was STATA 14.0 (StataCorp., College Station, TX, USA).

RESULTS
The sample consisted of 265 patients with CKD undergoing dialysis, of which 61.13% were from São Luís city and 38.87% from Recife city. Increased VAI levels prevalence was 58.49% and that of HWP was 29.81% (Table 1). Tables 1 and 2 show other information regarding sociodemographic, lifestyle, biochemical, clinical, and body composition aspects.
In the bivariate analysis according to sociodemographic and lifestyle variables (Table 3), VAI associated only with gender. Women had a greater prevalence of increased VAI levels above the cutoff points recommended by Amato and cols. (21) (p < 0.001). The HWP was associated with the variables gender, city, alcohol consumption, and smoking (p < 0.05), where patients with the presence of HWP were mostly female, living in Recife, consuming alcoholic beverages, and were smokers or ex-smokers (p < 0.05).   In the bivariate analysis, according to anthropometric, clinical and body composition characteristics (Table 4), VAI was associated with the variables %BF and TC, where patients with increased VAI levels presented such markers above the normal range (p < 0.05). The HWP, in turn, was associated with BMI, %BF, DM, SAH, and TC (p < 0.05). Most patients with HWP were overweight/obese, with a high % BF, diabetes, not hypertension and with a high CT.
In the adjusted analysis, VAI was associated with females with a high %BF (p < 0.05). The HWP was associated with female gender, alcohol consumption, overweight/obesity, high %BF, and low serum HDL-c levels (p < 0.05) ( Table 5).

DISCUSSION
In our study, we observed a high prevalence of increased VAI levels, which is higher than the HWP prevalence. The HWP prevalence observed was similar as that reported in the literature (12) and it stood out as a marker associated with more traditional factors of CR compared to VAI. Despite the high prevalence presented, VAI represented an indicator associated with fewer variables of CR evaluated in this study, such as alcohol consumption, smoking, DM, SAH, high LDL-c, and low HDL-c, in which there was no association, even analyzing more parameters together in its predictive equation.
It is important to highlight that most participants in our study were women and although we had not evaluated issues related to menopause, they were in an age group close to this period. As demonstrated by Amrita and cols. (23), this is a phase when many hormonal variations cause imbalance in oxidative processes, since estrogen exerts beneficial effects on endothelial dysfunction, modulating the lipid profile and increasing the production of nitric oxide. As hormone levels decrease, multiple metabolic changes occur, such as reduced glucose tolerance, dyslipidemia, redox state imbalance, changes in body fat distribution, hypertension, endothelial dysfunction, and vascular  inflammation, greatly contributing to increased cardiovascular risk (24). We found that gender associated with both HWP and VAI, so that being female increases by 1.46 and 1.80 times, respectively, these conditions. In this sense, the present study is a warning to the increased risk of CVD in female HD patients with CKD.
Evaluating from another aspect, the respective equations explain in part the difference observed between HWP and increased VAI levels prevalence. The VAI is calculated according to the formula proposed by Amato and cols. (6) which considers two more variables than the equation for calculating the HWP (BMI and HDL-c). Although the majority of patients with increased VAI levels had HDL-c levels within the recommended range, we observed that the BMI prevalence in the age group of overweight/obesity was higher among patients with increased VAI levels compared to those with a VAI within the normal range (p < 0.001), which could explain the fact that increased VAI levels prevalence is higher than that presented by the HWP in our study.
Developed by Lemieux and cols. (5) the HWP derives from the hypothesis that simple variables such as WC and fasting plasma TG concentrations could be used as screening tools for the identification of the atherogenic metabolic triad. This is the first study to evaluate HWP in patients undergoing dialysis treatment in Brazil and reports a 29.81% prevalence. Previous studies, such as of Freitas and cols. (12) which evaluated 15,105 civil servants, active and retired, aged between 35 and 74 years, of both genders, from six higher education institutions located in cities in different regions of Brazil, found a 24.6% prevalence. Cabral and cols. (25) evaluated 218 patients followed up by the Hypertensive and Diabetic Registration and Monitoring System Program (HiperDia) in two health units at São Luís in Maranhão, Brazil, and found a 33% prevalence. Oliveira and cols. (26) evaluated 191 individuals from the Bahia School of Nutrition and found a 20.2% HWP prevalence. The variations found in different HWP prevalence studies may be explained due to the use of different cutoff points for WC and TG levels, as well as ethnic differences and use of lipidlowering drugs (26).
Freitas and cols. (12), Cabral and cols. (25) and Oliveira and cols. (26) observed associations with older age, excessive alcohol consumption, ex-smoker, low HDL-c and LDL-c, TC, and C-reactive protein (PCR), fasting blood glucose ≥ 100 mg/dL or person with diabetes and presenting a greater number of CR factors. In the study by Zhe and cols. (9) with nondialysis renal patients, the concentrations of TG, TC, HDL-c, and LDL-c in the group with HWP were significantly higher than those in the group without the phenotype. In addition, the average intima thickness of the carotid artery, one of the most accepted substitute indexes for local and generalized atherosclerosis was the highest in the group with HWP. The authors point out that HWP can be useful to predict the risk of CVD in patients with CKD. In our study, we found similar associations, showing the importance of this marker for the identification of patients with pro-atherogenic characteristics, since CVD represents the main cause of death in patients with CKD (2).
We also observed an association between high levels of %BF and the studied visceral adiposity indicators. The literature reports that adiposity, especially visceral, plays an important role in the context of CVD because it produces higher levels of inflammatory cytokines. It is associated with insulin resistance, oxidative stress markers, and inflammation (27). Thus, our findings show the importance of assessing the indicators of visceral adiposity in patients with CKD undergoing HD.
The increased VAI levels, despite its high prevalence, was an indicator associated with fewer traditional CR factors when compared to HWP. It is worth mentioning that the equation and cutoff points proposed by Amato and cols. (21) for defining and classifying the VAI were validated for a population with different characteristics than those of the Brazilian one and did not involve specific subgroups, such as patients with CKD in HD.
Amato and cols. (28) in subsequent clarifications regarding the limitations of the use of VAI, reported that the use of this index is limited to non-Caucasian populations, since the process of modeling the VAI took place in a Caucasian population aged 19 and 83 years and BMI between 20 and 30 kg/m², which may cause adverse results when applied to other populations with different characteristics. In addition, the authors emphasize that, individually or in small samples, VAI should not be applied, especially in the presence of morbid obesity (BMI > 40 kg/m²), pendular abdomen, severe hypertriglyceridemia (>279 mg/dL) and/or use of fibrates. Although only 9.81% of patients had severe hypertriglyceridemia in our sample and none had morbid obesity, we did not investigate the use of fibrates, which may be considered a limitation in this study.
International studies  have sought to assess the predictive power of VAI to discriminate CR in the HD population. However, these studies were developed in the Asian population, which have important differences between health profiles and body composition observed in Brazil. In addition, the authors used cutoff points for WC, BMI, and age, which were different from those used in our study. The aspects mentioned may not have affected the performance of HWP, as the cutoff points used in our study are similar as those used in another study with a representative sample of the Brazilian population (12).
As study limitations, as already mentioned, we could not explore and compare gender-specific associations. In addition, as this is a cross-sectional study, it is not possible to establish a causal relationship. Another important aspect is that the data analyzed in that study come from two studies in different states, which used different sampling methodologies. We also do not rule out a possible selection bias, since in Recife (PE) data was collected by convenience. In addition, as it is a sample of the population in HD specifically of certain centers restricted to São Luís and Recife cities, we recommend caution in the extrapolation of our findings. In contrast, this is the first study to evaluate HWP and VAI among the population undergoing dialysis treatment in Brazil, and it has made important contributions to the knowledge of its prevalence, hitherto unknown, in patients with CKD on HD, showing that women have a higher risk and elucidating the association of a simple and low-cost HWP indicator with important traditional CR factors.
The findings of this study show a high prevalence of increased VAI levels and a prevalence similar as that reported in the literature regarding HWP. The women evaluated are at a higher risk of having both indicators of visceral adiposity altered. The HWP, despite having a lower prevalence than increased VAI levels does and although its equation evaluates fewer parameters together, is associated with traditional CR factors. This is a practical and low-cost tool that can be used for screening CR in renal patients under dialysis treatment in primary care. We reinforce that longitudinal studies are needed to validate the analyses interpreted here.