Relationship between Non-Alcoholic Fatty Liver Disease and Visceral Fat Measured by Imaging-Based Body Composition Analysis: A Systematic Review

: Imaging-based body composition analysis can quantify visceral fat, which is an important feature of lean non-alcoholic fatty liver disease (NAFLD) patients. This review assesses current evidence of the relationship between NAFLD, particularly hepatic steatosis, and visceral fat that is measured using imaging-based body composition analysis. PubMed Central and ScienceDirect were searched for studies that provided quantiﬁcation of the relationship between NAFLD, hepatic steatosis and visceral fat. Twenty studies comprising 15,763 subjects were included, consisting of the relationship with NAFLD ( n = 15) and the relationship with hepatic steatosis ( n = 7). All studies reported a positive relationship between NAFLD and visceral fat. For hepatic steatosis regardless of severity, only one study reported no correlation with visceral fat. Further results showed that visceral fat is more related to NAFLD and hepatic steatosis in females than males. More studies including NAFLD of different stages must be performed in the future to validate the degree of association between visceral fat and NAFLD at all stages as well as this relationship difference between genders.


Introduction
Non-alcoholic fatty liver disease (NAFLD) refers to a range of liver diseases that can be confirmed to not be caused by the overconsumption of alcohol [1].When fat infiltrates the liver and accounts for more than 5% of the liver's weight, it can be termed a fatty liver, also known as hepatic steatosis [2].NAFLD is an umbrella term that comprises a few different stages of liver conditions, beginning from hepatic steatosis to more severe or significant steatosis, followed by non-alcoholic steatohepatitis (NASH), then fibrosis and eventually cirrhosis.In hepatic steatosis, the accumulation of fat cells may result in the inflammation of the liver, known as NASH.As it worsens, fibrosis eventually occurs as the inflammation results in the scarring of the liver [3].
There is a changing trend of NAFLD as their prevalence and incidence are increasing over the years especially among different Asian countries, with the highest being reported as 43.3% in Shanghai, China as of 2016 [4].Despite this, in a global study with data collected from 102 countries, no country has a clear strategy for NAFLD, and only 32 countries have national NAFLD clinical guidelines [5].This poses a concern as there are different implications resulting from NAFLD.If left untreated, it can eventually lead to cirrhosis, which, although a later stage of NAFLD, can cause a nodular transformation of the liver, and the reversibility of the disease is less likely [3].It has been shown by many studies that NAFLD is associated with other illnesses including hepatocellular carcinoma, cholangiocarcinoma, gallstone disease, and coronary heart disease.NAFLD has been reported to have a greater risk of both all-cause and cardiovascular-related mortality [6].
NAFLD could be largely due to unhealthy diets such as overeating or overconsumption of fats and sugar, and hence it should be properly managed with the right nutrition therapy [7].Therefore, it is important to detect NAFLD at the early stage so that appropriate treatments and management, e.g., nutrition therapy, can be delivered to improve patient outcomes [8].Inevitably, the challenging part is that not all NAFLD patients are overweight or obese as approximately 10-20% of affected NAFLD patients are "lean NAFLD" [9], with some reported "lean NAFLD" comprising 7-20% of NAFLD patients among different studies [4].This calls for a comprehensive body composition assessment instead of relying on the traditional body weight to correlate with NAFLD in monitoring any anthropometric factors that can affect NAFLD progression.
With numerous factors that can lead to NAFLD, high visceral fat, also known as visceral adipose tissue (VAT), is one of the prominent observations present in a subset of NAFLD patients [10].It is also well established that beyond a certain threshold, excess VAT may result in a malfunction of adipose tissue and cardiometabolic disturbances [11].As such, VAT is a measurement of concern that can be studied for its relationship with NAFLD.Several ways can be used to measure VAT such as bioelectrical impedance analysis (BIA) and the various radiological modalities.Although easily accessible, multiple studies have found that BIA was inferior to radiological techniques for measuring VAT in terms of accuracy [12][13][14][15].Moreover, multidisciplinary collaboration engaging different expertise for healthcare delivery is essential to provide comprehensive care and management to NAFLD patients [16].
Radiological methods can accurately measure VAT regardless of the form of visceral fat area (VFA) or visceral fat thickness (VFT).Furthermore, scanning provides images of the liver for assessment of the severity.Although different imaging modalities present different advantages and disadvantages, ultrasound, magnetic resonance imaging (MRI), computed tomography (CT), and dual-energy x-ray absorptiometry (DEXA) can measure VAT as required [17].Many studies with a wide range of clinical interests have adopted the use of different imaging-based body composition analysis for VAT measurement, thereby proving its values and usefulness in clinical research [18][19][20][21].Considering the increasing prevalence of NAFLD, it is desirable to seek alternatives in the field of imaging-based body composition analysis since it can also help in assessing the degree of steatosis [22][23][24].
Despite the radiology technological advancements in assessing steatosis, liver biopsy remains the gold standard for diagnosing NAFLD [3].Thus, this systematic review aims to assess the existing evidence of the relationship between NAFLD and VAT that is measured by using imaging-based body composition analysis.Moreover, the degree of the association will be evaluated for its difference between genders.

Materials and Methods
This review and the study protocol were conducted in adherence to the Preferred Reporting Items for Systematic and Meta-Analyses (PRISMA) guidelines [25].

Study Selection
The inclusion criteria are: (1) The articles are published within 15 years from January 2008 to December 2022.(2) Quantification of the relationship between visceral fat with NAFLD was identified.(3) Visceral fat is measured directly by radiological technique.(4) The condition is not caused by excessive consumption of alcohol if specific liver conditions are reported instead of NAFLD.(5) Provides r, odds ratio (OR), risk, hazards ratio (HR) or coefficients together with 95% confidence interval (CI) for quantification of the relationship.
The exclusion criteria are: (1) Articles were published more than 15 years ago, i.e., before January 2008.(2) No correlation, association nor quantification of association between visceral fat with NAFLD was identified.(3) Visceral fat measured from non-radiological techniques, e.g., BIA. ( 4) No mention of alcoholism being excluded if specific liver condition is reported instead of NAFLD.(5) Outcomes not appropriate.(6) Conference abstracts or reviews.

Quality Assessment and Data Extraction
Studies that meet the inclusion criteria were then subjected to quality assessment using the mixed-methods appraisal tool (MMAT) checklist [26].The MMAT allows for the screening of various study designs including qualitative, quantitative, and mixed-method studies [27].This comprises 5 different categories of the design study, namely (1) qualitative, (2) non-randomized, (3) randomized controlled trial, (4) quantitative descriptive, and (5) mixed-methods study.Following 2 screening questions, within each category, the quality of the study was based on 5 core criteria, whereby, within each criterion, there are three levels of score, namely "Yes", "No" and "Can't tell".The MMAT score was not used to eliminate studies, but rather to signify the studies' methodological solidity, with the following approach: 4-5 criteria met = high quality, 2-3 criteria met = moderate quality, and 0-1 criteria met = low quality.
Thereafter, descriptive characteristics of the studies are extracted as follows: first author with the year of publication, journal title, the aim of the study, study design, the MMAT score, population size and their breakdown by gender, age, and location of the study conducted.Details about their radiological procedure were also extracted independently, namely modality used in assessing VAT and liver (if applicable), equipment model and/or brand as well as the landmark used in measuring VAT.Finally, their statistical analysis method, association and correlation results, as well as conclusion were then extracted.Numerous forest plots were plotted for studies with odds ratio (OR) only just for a clearer comparison and visualization of the results.Only 1 study [28] with OR was not included in any of the forest plots due to the different parameter of liver-spleen attenuation index (LAI) being used to diagnose NAFLD.
The descriptive characteristics of the studies together with their MMAT score are shown in Table 1.Out of the 20 studies, only 16 reported their study design.There were reported seven cross-sectional studies, three prospective studies, two retrospective studies, one cohort study, one large cohort longitudinal prospective study, one population-based cohort, retrospective study, and one cross-sectional, retrospective cohort study.Except for two studies which were conducted on participants aged between 12 and 18 [36] and 15 to 19 [33], respectively, the rest of the studies were conducted on adults.Altogether, two studies were conducted in the USA, two studies in Brazil, one in France, three in Japan, three in China, and nine in Republic of Korea.Studies where the design of the study was not reported were not subjected to MMAT score quality assessment to avoid an inaccurate MMAT score overall.Collectively, these 20 studies included 15,763 subjects comprising 7808 males and 7955 females.One study recruited only males [36] and had the lowest number of participants (n = 57), while the study with the highest number of participants had a total of 3846 [47].Furthermore, the risk of bias for several components in the study design was assessed individually for all included studies using the Risk of Bias in Non-randomized Studies of Exposure (ROBINS-E) tool (Figure 2) [48].
The imaging modalities and tests used in assessing VAT and liver profile, respectively, are collated in Table 2. CT (n = 13) was the most common method for measuring VAT, followed by ultrasound (n = 4) and MRI (n = 3).In assessing the liver, two studies used ultrasound-guided percutaneous liver biopsy, four studies used MRI of which three performed MRS, one study used NAFLD-LFS score, which was developed from MRS, four studies used CT, and nine studies used ultrasound of which one performed a fibro scan.
No study meeting inclusion criteria that assess the relationship between NAFLD and VAT measured by using DEXA was identified.Main relationship findings for all studies are summarized in Table 3.It should be noted that while some studies related VAT to NAFLD in general, hepatic steatosis, severe or significant steatosis were assessed individually in a few studies as they are part of NAFLD.It should also be noted that in summarizing these findings, association with any later stages of NAFLD including NASH and fibrosis was not reported or determined in all of the studies.Only one study [44] also included for only with fibrosis but the association between visceral fat and fibrosis was also excluded from Table 3 due to the limitation in the number of studies assessing for it.As seen from Table 3, 13 studies assessed the relationship between visceral fat with NAFLD of which 1 study used LAI, which is an indicator of NAFLD, 3 studies between visceral fat with hepatic steatosis, 1 study between visceral fat with significant hepatic steatosis, 1 study between visceral fat with both mild and severe hepatic steatosis separately, 1 study between visceral fat with both NAFLD and hepatic steatosis separately, and 1 study between visceral fat with both NAFLD and severe hepatic steatosis separately.In the logistic regression analysis, visceral fat was found to be one of the independent predictors of NAFLD (β-coefficient: 1.606, p = 0.014).
In the correlation, visceral-to-subcutaneous ratio has r = 0.771 with non-NAFLD but presented with higher r = 0.790 with NAFLD patients.
Visceral fat cut-off point for the prediction of NAFLD adolescents in both genders was reported.However, the cut-off point cannot be generalized to other populations.
Visceral adiposity was positively correlated with the degree of hepatic steatosis in overweight or obese potential living liver donors.Additionally, visceral adiposity may be an independent risk factor for overweight or obese NAFLD.

Hepatic steatosis
In men, the degree of hepatic steatosis was positively correlated with VFA (r = 0.307; p < 0.001).
In women, the degree of hepatic steatosis was positively correlated with only VFA (r = 0.387; p < 0.001).

The Relationship between VAT with NAFLD
Altogether, there were 15 studies directly assessing the relationship between visceral fat with NAFLD of which 1 study was assessed with LAI, where an LAI of lower than 0.9 is diagnosed as NAFLD.Of these 15 studies, 1 study has a β-coefficient, 2 studies have HR, 4 studies have both OR and a correlation coefficient, and 8 studies have only OR as their outcome.
For studies using OR, VAT was indicated to be independently associated with NAFLD (Figure 3), presenting statistically significant results after analysis and adjusted for variables.Despite the different study designs and subjects' profiles, this association was consistently low with four out of five studies having a result of OR < 2.0 while the highest reported result was OR 2.53 (95% 2.04-3.12,p < 0.001).In another study [28] included in this review separately from Figure 3, where it was not included in the forest plot as it uses LAI, it was found that even without obesity but with increased VAT presented with OR 5.88 (95% CI 1.03-33.52,p = 0.046) for developing NAFLD.Surprisingly, patients with high BMI but low VFA did not display a greater risk of lower LAI.[30,36,39,42,43].Some studies further broke down the association by genders.As shown in Figure 4, three out of four studies showed higher ORs for females than males, signifying that females with high VAT were at greater risk of developing NAFLD than males.
Two studies [40,47] were included separately in this review as another forest plot (Figure 5) because their OR was derived differently by analyzing between those subjects in the highest VAT category and those subjects in the lowest VAT category, differentiated for between both genders.Results from both studies concurred that females are at greater risk of being affected by NAFLD than males despite having a significantly lower VAT.While Kure et al. (2019) [40] showed OR 21.6 (95% CI 11.3-41.2,p < 0.01) of developing NAFLD for females in comparison to males with OR 8.20 (95% CI 5.38-12.9,p < 0.01), Baek et al. (2020) [47] showed OR 17.84 (95% CI 7.12-44.71,p < 0.001) for females and OR 6.14 (95% 2.96-12.73,p < 0.001) for males of developing NAFLD.

The Relationship between VAT with Hepatic Steatosis
Conjointly, there were seven studies assessing with hepatic steatosis of all stages.One study assessed the correlation between VAT with both mild and severe hepatic steatosis using HU, whereby the lower the HU value, the denser the visceral fat tissue and the more severe the steatosis is.The reported correlation coefficient was r = −0.481(p < 0.001) [41], indicating that increasing VAT has a positive correlation with steatosis of increased severity Three studies [32,37,44] assessed the correlation between VAT and specifically hepatic steatosis with one study [37] differentiated between genders.The overall correlation coefficient reported was r = 0.307 (p = 0.014) [32] and r = 0.569 (p < 0.001) [37].Only one study established that they are not correlated, presenting a statistically insignificant and lower value result of r = 0.13 (p = 0.090) [44].
When divided by gender, it was found in [37] that the correlation for developing hepatic steatosis from increased VAT was greater in males (r = 0.576, p < 0.001) than females (r = 0.473, p = 0.003).Contrarily, in another study [46] that assessed the correlation of VAT with the degree of hepatic steatosis differentiated between genders on top of NAFLD, it was determined that the correlation between the degree of hepatic steatosis with VAT was greater in females with r = 0.387 (p < 0.001) than males with a correlation of r = 0.307 (p < 0.001).
Two studies assessing the association between VAT with specifically severe or significant hepatic steatosis present with the results of OR 1.010 (95% CI 1.001-1.019,p = 0.028) [35] and OR 5.08 (95% CI 2.00-12.88,p = 0.0006 [39], respectively.When further divided by gender, one study [35] established that in males, VAT was the sole independent risk factor for significant hepatic steatosis with OR 1.008 (95% CI 1.001-1.011,p = 0.045), whereas in females, it was more closely associated and was just one of the independent risk factors with OR 1.029 (95% CI, 1.010-1.048,p = 0.002).

Discussion
From the results, it can be ascertained that NAFLD and degree of hepatic steatosis were related to increased visceral adiposity.Besides the study by Kang et al. [44] which showed statistically insignificant correlations for only hepatic steatosis, all other studies demonstrated a statistically significant positive association and correlation between VAT and NAFLD, which can be attributed to a few reasons.
The liver regulates energy by receiving blood from the gut through the portal vein after food intake and will be stored within until being utilized.Visceral fat is metabolically more active as it will produce signals to regulate energy, so it is susceptible to overwork.This will then lead to inadequate blood and subsequent inadequate oxygen supply, which will then cause the fat cells to be swollen, resulting in damaged fat tissues that will then supply harmful substances into the blood to harm the liver [3].This means that visceral fat can let out free fatty acids (FFA) and triglycerides to circulate, which can cause fatty acids to accumulate straight into the liver through portal circulation [49].
The physiology involves an increase in the inflammation markers and immune cell activity within liver tissue which will then decrease the adiponectin, an insulin sensitivity hormone.Eventually, hepatic insulin clearance will be decreased, increasing hepatic fat deposition.Increased FFAs in the liver will also bring about the production of cytokine, which can contribute to the inflammation pathways associated with NAFLD [49].This is also as shown in a study performed by Denkmayr et al. (2018) [50] that lean NAFLD patients who were characterized as having higher VAT degrees, also had higher adipokine, which is a cytokine, than lean healthy patients.Moreover, with adiponectin being an adipocyte-derived hormone with properties insulin-sensitizing and anti-inflammation, decreased levels of adiponectin may lower the ability to protect from NAFLD [51].Besides, in another study [52], it was identified that VFA was dose dependently and independently associated with alanine aminotransferase (ALT), which is an enzyme that can be a liver damage marker.
As liver disease progresses, there would be an accumulation of hepatic fat deposits, which could explain the positive relationship between visceral fat with severe or significant steatosis.As above mentioned, only one study [44] out of all included studies also reported for the presence of fibrosis and no other included studies reported any presence of later stages of NAFLD.As such, the stages of NAFLD the patients included in the studies were unknown.In this study [44], high visceral fat index was found to be one of the independent risk factors for advanced fibrosis.In another instance, also in the abovementioned study by Denkmayr et al. [50], it unexpectedly showed that even lean NAFLD patients with normal BMI but a higher VAT degree, have a higher rate of fibrosis than overweight and obese NAFLD patients.In other separate studies not included in this review, VAT is also found to be positively associated with fibrosis in NAFLD [53,54].These results could signify presence of an underlying mechanism between visceral fat with the later stages of NAFLD and may not be limited to only fibrosis.As such, this unascertained relationship between visceral fat with later stages of NAFLD could be confounding factors in affecting the association results and ought to be accounted for.
Vitamin D deficiency could be another possible cause for the relationship between VAT and NAFLD [55].A few studies have shown that there was a positive association between concentrations of vitamin D and adiponectin [56,57].Not only that, in a 12-month longitudinal follow-up study [57] within these studies, it was displayed that an increased vitamin D level results in a decreased secretion of several inflammatory cytokines.While one study [58] concluded that low vitamin D was closely associated with NAFLD regardless of the visceral obesity status, other studies [59,60] have established that there is a significant relationship between vitamin D status and visceral fat quantity with lower vitamin D status having a higher amount of visceral fat.Separately, Choi et al. [59] have also demonstrated that vitamin D deficiency has a higher NAFLD prevalence.Taken together, these matched findings suggest that low levels of vitamin D can contribute to higher visceral fat and result in NAFLD, considering how a significant reduction in adiponectin and increase in cytokines was positively associated with NAFLD as mentioned [51].
Results divided by gender across all studies were summarized in Table 4.When differentiated by the genders, there were conflicting results.Current evidence suggests that males are more prone to develop higher VAT than females due to their underlying mechanism of higher and larger chylomicrons [61].Taking this into account, males should have a greater risk of developing NAFLD given their higher visceral fat.Unexpectedly, more results from this review show that without considering other factors, VAT individually as an independent variable is more related to NAFLD for females than males.As seen in Table 4, five out of six studies indicated females with high VAT have higher ORs in developing NAFLD than males with high VAT.For the relationship with hepatic steatosis, one study deemed the correlation was higher for males while another two studies ascertained it will affect females more.While the reason cannot be validated for the only two studies indicating NAFLD and steatosis are more likely to affect males than females, this illustrated the complex nature of the relationship difference for both genders.Notwithstanding the fact that visceral fat was analyzed independently to give the results in Table 4, a separate meta-analysis [62] performed has determined that while increased serum total testosterone in males decreased the risk of NAFLD, it increases NAFLD risk in females.The difference in the effect of these sex hormones have on the results, thereupon reinforced the difference in the relationship with NAFLD between genders.
Even though visceral fat was analyzed independently to derive results shown in Table 4, it should be known that there is a higher risk of developing NAFLD in postmenopausal women due to their decreased level of estrogen, which affected the visceral distribution and liver lipid metabolism, leading to dyslipidemia milieu [63].A populationbased cross-sectional has demonstrated that NAFLD was more possible to affect women than men, especially in those aged 56-60, i.e., the post-menopausal period [64].One of the studies included in this review [47] also distinctively showed there were more NAFLD female subjects (n = 326) from age 50 onwards than the younger ones (below age 50, n = 52), as opposed to male subjects where the distribution of NAFLD patients was more evenly distributed across all ages.This indicates that age which impacts the menopausal status, can concurrently affect the NAFLD epidemiology and pathophysiology [65].
It is noteworthy to point out that 5 of the 20 studies also involve the use of visceralto-subcutaneous ratio (VSR) as a parameter to relate with NAFLD and steatosis.One study [34] presented with a result of HR 1.51, 95% CI 1.03-2.19,p = 0.033 to develop NAFLD.In another study [41], it showed that there was a higher VSR in moderate/severe than mild steatosis but is not statistically significant (p = 0.258).Lean NAFLD has the highest VSR in one study [43] but is also statistically insignificant (p = 0.127).However, in this study, VSR showed a significant association with fatty liver disease (r = 0.342, p < 0.001).There also exist a positive correlation in another study [33] with a result of r = 0.790.When divided by gender, one study [47] showed r = 0.031, p < 0.001 for men; r = 0.162, p < 0.001 for women, as well as OR 1.99 95% CI 1.22-3.26;p = 0.003 for men, OR 6.73 95% 4.10-11.03;p < 0.001 for women.These results are comparable to those from visceral fat and, therefore, suggest that they may not be of less importance than using only visceral fat as a parameter.
To the best of our knowledge, this is the first systematic review assessing the relationship between NAFLD and visceral fat measured using imaging-based body composition analysis.We acknowledge that this review has its own set of challenges and limitations.Meta-analysis was not performed as the included studies and results were too heterogeneous.There are differences in the study design and the included subjects' profiles differ between all with some of the studies recruiting subjects with existing conditions including diabetes and Crohn's disease.It should also be noted that the association result shown for the measurement unit of the VAT increase was not constant.While one study showed the result for a per-cm increase in VAT [29], some showed a cm 2 increase [40,47], some showed a 1-SD increase in volume (L) [43], and some calculated into and used the VFA index as per 1.0 cm 2 /m 2 increase [45].However, VAT was analyzed independently of its association with NAFLD in these studies, and the consistency of the results proved the reproducibility of the positive association between VAT and NAFLD regardless of the measurement unit.
Particularly, there was a challenge in sourcing for studies performed in multiple countries from different continents to make this review more generalizable.Of the 20 studies, only 5 studies were from non-Asian populations.This calls for the need to have more studies being performed to account to a wider demographic population.Nevertheless, all the remaining 15 studies were from east Asian countries.Thus, so far as possible, this review provides a comprehensive coverage to be representative of the east Asian population.
In this review, there were different imaging modalities and landmarks used in measuring VAT.Nonetheless, the lack of sufficient studies makes this standardization challenging and the consistent association and correlation results rationalized the importance and validity of the different imaging-based body composition analysis in measuring body fat.This is crucial as there are no single imaging modalities that will fit all patients due to their different risks and subjects' suitability.

Conclusions
Without considering other factors, more results from this review showed that high VAT as an independent factor leads to NAFLD and steatosis affecting females more than males.The large variance, as seen in the data collected in this review, clearly shows the need for more clinical studies to be performed in the future.Nevertheless, a key takeaway from the result is that VAT measured from imaging-based body composition analysis is closely related to NAFLD and hepatic steatosis.This can reinforce the existing evidence that regression or prevention of NAFLD can happen by aiming to reduce the VAT as part of nutrition therapy or management.However, the degree of the association and correlation as well as these results between genders vary between different studies.More studies with both genders, possibly including various age groups and NAFLD stages, still need to be performed to establish the degree of association between visceral fat and all stages of NAFLD and the difference in this association between genders, as well as to further elucidate how this changes with age.

Figure 1 .
Figure 1.Flow diagram of studies retrieval.

3. 5
MHz multifrequency transducer (broad band) Distance between the internal face of the same muscle and the anterior wall of the aorta.Ultrasound 3.5 MHz multifrequency transducer (broad band)

Figure 5 .
Figure 5. Included studies with OR for developing NAFLD between subjects of largest VFA and subjects of lowest VFA for both genders [40,47].

Table 1 .
Descriptive Characteristics of Included Studies.
[46].7 (N.R.) Kyoto, JapanLee et al. (2022)[46]Visceral fat area is an independent risk factor for overweight or obese non-alcoholic fatty liver disease in potential living liver donors Evaluate the correlation between hepatic steatosis, determined by biopsy, and visceral adiposity, measured by CT, in overweight or obese potential living liver donors, and to investigate the risk factors for overweight or obese NAFLD.N.R. indicates not reported, N.A. indicates not applicable.

Table 2 .
Radiological modality details of the included studies.

(Year) Visceral Fat Measurement Modality Machine Model and/or Brand Landmark for Visceral Measurement Liver Feature Assessment Method Machine Model and/or Brand
[29]so et al. (2008)[29]Ultrasound

Table 3 .
Findings from the included studies.

Table 4 .
Summary of association and correlation results differentiated by genders.