Introduction

Non-alcoholic fatty liver disease (NAFLD) is an abnormal deposition of fat in the liver without alcohol consumption or other specific causes. It is the most prevalent liver disease worldwide. NAFLD is strongly associated with obesity, hypertension, dyslipidemia, cardiovascular disease, impaired glucose tolerance, and type 2 diabetes (T2D). In addition, other factors, such as genetic and environmental, are also associated with disease progression [1].

Individuals with a BMI within normal clinical standards may have increased levels of visceral adiposity, which is a predictor for developing metabolic syndrome (MetS), characterized by a set of factors that increase cardiometabolic risk [2,3,4,5,6,7,8]. Visceral adipose tissue releases pro-inflammatory and pro-fibrogenic mediators associated with oxidative stress. Body fat compartments have differential effects on metabolic regulation, which may be related either to organ damage or to mitigation of harmful effects. In 1947, Vague firstly described the notion that certain patterns of fat distribution, in particular “central” obesity, predisposed to diabetes, atherosclerosis, gout, and urinary stones and that such fat distribution varied according to gender. Vague coined the terms “android obesity” (predominantly male, fat accumulated mostly in the trunk) and “gynecoid obesity” (predominantly female, whose sites of accumulation in hips and thighs) [9].

Thus, BMI is a useful but incomplete form of assessing nutritional status, since it cannot detect less generalized fat distribution patterns. Anthropometric measures that assess the distribution of body fat seem to better predict cardiovascular risk [10, 11]. Current evidence indicates that waist circumference (WC) more efficiently detects central adiposity and its association with co-morbidities [12,13,14,15]. Neck circumference (NC) is also used as an anthropometric marker of overweight and obesity, in addition to an association with central adiposity [13, 16, 17]. Hip circumference (HC) seems to be associated with better cardiometabolic outcomes because it is linked with the muscle mass of gluteus and quadriceps and favors insulin sensitivity [5, 6, 18]. Waist-to-hip ratio (WHR) is also a measurement that assesses central obesity and its association with co-morbidities. WHR is calculated by the simple division of WC by HC [16, 19].

The hypothesis of this study is that NAFLD in individuals with obesity may be significantly associated with WC, NC, HC, and WHR. This study aims at analyzing the existence of correlations between these anthropometric parameters and the occurrence and intensity of liver steatosis, steatohepatitis, and fibrosis assessed by histopathological examination in individuals with obesity.

Methods

Study Design

This is a cross-sectional study carried out in a tertiary university hospital in 2019. The research protocol was evaluated and approved by the local institutional review board under the opinion: 3.512.406/UNICAMP (CAAE 16562219.0.0000.5404).

Study Population and Sampling Calculation

This study included individuals of any gender, aged 18 to 70 years old, with BMI ≥ 35 kg/m2, which underwent bariatric surgery. Exclusion criteria were previous or current liver disease of other cause, serology abnormalities for viral hepatitis, previous or current use of alcohol, current or recent use of hepatotoxic medications, and previous biliary obstruction.

A sample calculation for cross-sectional studies was carried out considering the following parameters: total population of 150 individuals (standardized volume of individuals undergoing bariatric surgery in the current facility each year), 70% proportion of NAFLD patients, precision of 5%, and 95% confidence interval. A minimum sample of 103 participants was estimated.

After applying the inclusion and exclusion criteria, from a total population of 132 individuals, 119 participants were selected. All individuals included in this research underwent open Roux-en-Y gastric bypass surgery. Fig. 1 shows the flowchart of the study population.

Fig. 1
figure 1

Flowchart of the study population

Full size image

Liver Biopsy

Wedge liver biopsy was performed in all individuals included in the study, through the extraction of a 2-cm fragment from segment III or IV.

Study Variables

Demographic, clinical, anthropometric, laboratory, and histopathological variables were analyzed. The demographic and clinical variables analyzed were: age (expressed in years); gender (male or female), and presence of co-morbidities (hypertension, T2D, and dyslipidemias).

The anthropometric variables analyzed were: BMI (kg/m2); WC (cm); HC (cm); NC (cm), and WHR. Circumferences were measured by members of the research team the day before surgery using a standardized technique and the same measuring tape.

Histopathological variables were classified according to the Brazilian Society of Hepatology criteria, which analyzes the parameters from a semi-quantitative and topological basis, using the criteria of Kleiner and collaborators [20, 21]. Histological parameters were classified as follows: macrovesicular and/or microvesicular steatosis (classified ordinally as absent (0), mild (1), moderate (2), and severe (3) and dichotomously as absent or present); steatohepatitis analyzed through the presence of acinar and/or portal inflammation (ordinally classified as absent (0), mild (1), moderate (2), and severe (3) and dichotomously classified as absent or present), and perisinusoidal and/or periportal fibrosis (classified ordinally as absent (0), mild (1), moderate (2), and severe (3) and dichotomously as absent or present).

Statistical Analysis

For the comparison of proportions, the chi-square test or Fisher’s exact test was used, when necessary. To compare continuous variables, the Mann-Whitney test was used. The analyses of correlations between continuous and/or ordinal variables were performed through simple and multiple linear regressions. Considering the disproportionately higher number of females in the current sample, subset analyses excluding male individuals were also performed. The level of significance used for the statistical tests will be 5% (p < 0.05). To perform the analyses, the computer program SAS System for Windows (Statistical Analysis System), version 9.2; SAS Institute Inc., 2002–2008, Cary, NC, USA, was used.

Results

Of the 119 individuals included, 105 (88.2%) were female. The mean age was 38.8 ± 9.3 years and the mean BMI was 37.6 ± 3.1 kg/m2. The prevalence of NAFLD histopathological aspects was: steatosis (91 individuals; 76.5%); steatohepatitis (59; 49.6%), and liver fibrosis (61; 51.3%). In regard to co-morbidities, it was observed that 55 individuals (46.2%) had hypertension, 18 (15.1%) presented T2D, and 27 (22.7%) had dyslipidemia.

Comparing individuals with or without steatosis, WC was significantly higher in individuals with steatosis (103.5 ± 9.9 vs. 99.4 ± 8.4; p = 0.03). Excluding males, individuals with steatosis had significantly higher WC (102.2 ± 9.6 vs. 98.3 ± 8; p = 0.04), triglycerides (117 ± 52.6 vs. 90 ± 47.5; p = 0.04), and total cholesterol (179.2 ± 43.7 vs. 153.6 ± 46.5; p = 0.03). There were no statistically significant differences regarding other variables (Table 1).

Table 1 Comparison between individuals with and without liver steatosis
Full size table

It was observed that individuals with steatohepatitis presented significantly higher BMI (38.2 ± 3.2 vs. 36.7 ± 2.8; p = 0.01), WC (105.3 ± 10.4 vs. 99.6 ± 8.8; p = 0.002), and WHR (1 ± 0.1 vs. 0.9 ± 0.1; p = 0.02) than individuals without steatohepatitis. Excluding males, BMI (38.2 ± 3.4 vs. 36.5 ± 2.8; p = 0.007) and WC (102.9 ± 9.6 vs. 98.7 ± 8.7; p = 0.02) were significantly higher in individuals with steatohepatitis. There were no significant differences in regard to other variables (Table 2).

Table 2 Comparison between individuals with and without steatohepatitis
Full size table

As for liver fibrosis, it was observed that age (40.6 ± 9.7 vs. 37 ± 8.5; p = 0.03) and levels of hemoglobin A1c (6.5 ± 0.5 vs. 5.6 ± 0.5; p = 0.004) were significantly higher among individuals with fibrosis, with no significant differences in regard to other variables. Excluding males, ALT levels were significantly higher in individuals with fibrosis (24.3 ± 24.8 vs. 18.8 ± 8.9; p = 0.03) (Table 3).

Table 3 Comparison between individuals with and without liver fibrosis
Full size table

Univariate correlation analysis was performed between the intensity of liver steatosis (outcome) and the study variables, and a statistically significant positive correlation was observed with WHR (R = 0.2; p = 0.04). Excluding males, WC was significantly and positively correlated with the intensity of steatosis (R = 0.2; p = 0.02); there was also a marginally significant and positive correlation between WHR and the intensity of steatosis (R = 0.2; p = 0.05). Table 4 presents the complete correlation analysis. After a multivariate analysis including significant or marginally significant variables associated with the intensity of steatosis in females, WC was independently associated with this outcome (R = 0.02; p = 0.02) (Table 5). Univariate correlation analysis between steatohepatitis intensity (outcome) and study variables revealed significant positive correlations between BMI (R = 0.2; p = 0.02), NC (R = 0.2; p = 0.03), WC (R = 0.2; p = 0.006), and blood glucose (R = 0.3; p = 0.005). Excluding males, BMI was independently associated with the intensity of steatohepatitis (R = 0.3; p = 0.006). Table 6 shows the complete univariate simple regression analysis for the outcome of steatohepatitis intensity. After identifying the significant variables in the univariate analysis, multivariate analysis was performed using multiple regression considering the entire study population, where it was observed that BMI (R = 0.2; p = 0.02) and blood glucose (R = 0.2; p = 0.009) were independently correlated with steatohepatitis intensity (Table 7). Upon univariate correlation analysis between the intensity of liver fibrosis (outcome) and the study variables, significant correlations were observed with age (R = 0.3; p = 0.004), blood glucose (R = 0.2; p = 0.03), and hemoglobin A1c (R = 0.2; p = 0.04). Excluding males, no significant variable was identified. Table 8 shows the complete univariate simple regression analysis for the outcome of fibrosis intensity. After identifying the significant variables in the univariate analysis considering the entire study population, multivariate analysis was performed using multiple regression, where it was observed that age (R = 0.3; p = 0.04) was independently correlated with the intensity of fibrosis (Table 9).

Table 4 Univariate correlation analysis (simple regression) between liver steatosis intensity (outcome) and study variables
Full size table
Table 5 Multivariate correlation analysis (multiple regression) between steatosis intensity (outcome) and significant or marginally significant variables in univariate analysis after exclusion of male individuals
Full size table
Table 6 Univariate correlation analysis (simple regression) between steatohepatitis intensity (outcome) and study variables
Full size table
Table 7 Multivariate correlation analysis (multiple regression) between steatohepatitis intensity (outcome) and significant variables in univariate analysis (overall population)
Full size table
Table 8 Univariate correlation analysis (simple regression) between liver fibrosis intensity (outcome) and study variables
Full size table
Table 9 Multivariate correlation analysis (multiple regression) between liver fibrosis intensity (outcome) and significant or marginally significant variables in univariate analysis (overall population)
Full size table

Discussion

Among the main findings of the present study, the significance of both WC and WHR as markers of steatosis stands out. Both are indirect anthropometric markers of body fat distribution usually correlated with the volume of visceral fat [17, 22]. BMI, the most widely used anthropometric marker in daily clinical practice, was significant only among individuals with steatohepatitis. Among individuals with hepatic fibrosis, there were no anthropometric markers that differed significantly; age and hemoglobin A1c were significantly higher among these individuals, indicating that the duration of the disease and the degree of insulin resistance are more important than the anthropometric parameters at this stage of the disease. The present findings confirm the precedence of WC and WHR as reliable markers of visceral adiposity, and this is also true in the assessment of liver disease [23]. In addition, there was a significant and independent positive correlation between WHR and steatosis intensity, indicating that the higher the WHR, the greater the severity of steatosis. Likewise, BMI and glucose showed positive and independent correlations with the intensity of the steatohepatitis, indicating that the higher both parameters, the greater the severity of the steatohepatitis. Fibrosis, on the other hand, showed a significantly positive and independent correlation with age, indicating that in this population, the older the individual, the greater the degree of liver fibrosis. NC was not significantly associated or independently correlated with any of the histopathological aspects of NAFLD analyzed in the study population.

According to studies carried out by Atri et al. [24] in a population of 106 Indian women with morbid obesity, in which 73.6% had NAFLD diagnosed by ultrasound, WC, BMI, and waist-to-height ratio were significantly associated with steatosis. Another relevant finding of the present study was that HC did not prove to be an isolated significant anthropometric protective factor for steatosis, steatohepatitis, or fibrosis. Katz et al. [25] previously observed that, among 1776 Chinese women, HC had an inverse and strong relationship with high levels of triglycerides and low levels of HDL. Similarly, Snijder et al. [26] observed, in a study which enrolled 11,247 participants, that HC had an inverse and independent relationship with the presence of diabetes, hypertension, and dyslipidemia. On the other hand, the current study did not demonstrate negative correlations, that is, protection between HC and aspects of the NAFLD spectrum. It is likely that the results presented by the aforementioned authors may be reproducible in more heterogeneous populations, composed of individuals with or without obesity, whereas in the current study, whose population only comprised individuals with severe obesity, this association could not be observed. Another hypothesis is that the metabolic protection provided by HC in such studies does not extend to NAFLD, being limited to other components of cardiometabolic risk. Further research is needed to reach more comprehensive conclusions in this regard.

NC, defended by some researchers as so significant as WC as an anthropometric marker of metabolic disease [27, 28], did not show associations with NAFLD in the current study. In a study by Borel et al. [22] which enrolled 305 French adult women with severe obesity, it was observed that NC was a clinically relevant anthropometric predictor of central obesity. In another study by Huang et al. [27], in an adult Chinese population composed of 4053 individuals, NC was associated with a higher prevalence of NAFLD diagnosed by ultrasound scan, being considered by the authors as a valuable predictor of NAFLD, especially when analyzed in conjunction with other anthropometric measurements, such as WC and HC. For Li et al. [28], NC was an independent predictor of NAFLD diagnosed by ultrasound scan in lean men in a population of 2668 adults in Shanghai in Eastern China. According to the study by Jian et al. [29] in a sample of 1698 Chinese adults, it was observed that NC is an effective supplementary parameter for screening NAFLD assessed by ultrasound scan; there was a significant and positive correlation between NC and liver fat content. The findings of the present study do not corroborate these results. It must be considered that such results refer to studies carried out in populations of varying ethnicities, so that they are likely not entirely reproducible in the Brazilian population. Furthermore, the current study used histopathological examinations for the diagnosis of NAFLD and its aspects, while the aforementioned studies mostly used imaging methods for this purpose.

In multivariate analyses, it was shown that BMI and hemoglobin A1c were markers independently correlated with the intensity of steatohepatitis, i.e., the greater their values, the more intense the inflammation, emphasizing the importance of glucose metabolism on steatohepatitis. In a literature review carried out by Wattacheril et al. [30] and in the study carried out by Pappachan et al. [31], the various pathophysiological pathways that are involved in the progression and pathogenesis of steatohepatitis are shown, among them, the pathway of abnormal glucose metabolism and insulin [32].

Currently, there is no consensus on the use of anthropometric parameters for screening and diagnosis of NAFLD. These divergences have a multifactorial origin in relation to these parameters, which may be related to factors such as ethnicity, age, gender, genetic, and socioeconomic aspects, as well as varying samples and methodology issues [27,28,29, 31, 33, 34]. Despite these controversies, it is possible, through the findings of the present study, to postulate that the measurement of WC and WHR are simple and easily reproducible ways of screening individuals with risk factors for NAFLD and NAFLD. The adoption of this measurement as a routine in daily outpatient care of the bariatric population could be used as an instrument to indicate the need for further laboratory and imaging tests in order to diagnose and perform the risk gradation of severe forms for NAFLD. In addition, it could establish criteria for greater metabolic risk during the discussion on treatment options for these individuals, particularly with regard to the indication of bariatric/metabolic surgery, especially a greater emphasis on the importance of WC and WHR over BMI alone [3, 6, 18, 35, 36]. Considering the difficulty to detect NAFLD by visual inspection during surgical procedures, these parameters may also help to select individuals with more necessity of intraoperative liver biopsy during bariatric surgery [37,38,39].

The present study has limitations that must be considered. The population was exclusively composed of individuals with obesity undergoing bariatric surgery, thus it is considerably homogeneous and does not allow a prompt extrapolation of the findings to more heterogeneous groups. Nevertheless, the histopathological assessment of NAFLD is an extremely relevant strength of this research, providing a nuanced analysis of the disease spectrum, which is not possible and neither so accurate in studies based on imaging methods.

Conclusions

Anthropometric parameters were significantly associated with the occurrence of NAFLD histopathological aspects in individuals undergoing bariatric surgery. WC and WHR were associated with the occurrence of steatosis; BMI, WC, and WHR were associated with the occurrence of steatohepatitis. A significant positive correlation was identified between WHR and steatosis intensity.