VAT fat is bad for the liver, SAT fat is not!


  • Jian-Gao Fan,

    1. Center for Fatty Liver Disease, Shanghai First People's Hospital, Jiaotong University, Shanghai, China; and
    2. Australian National University Medical School at The Canberra Hospital, Canberra, Australian Capital Territory, Australia
    Search for more papers by this author
  • Geoffrey C Farrell

    Corresponding author
    1. Australian National University Medical School at The Canberra Hospital, Canberra, Australian Capital Territory, Australia
    Search for more papers by this author

Professor Geoff Farrell, The Canberra Hospital and ANU Medical School, Garran, ACT 2604, Australia. Email:

Obesity predisposes to several serious medical conditions, including the spectrum of liver abnormalities collectively referred to as non-alcoholic fatty liver disease (NAFLD).1–5 The global obesity pandemic so evident in the Asia–Pacific region underlies a significant increase of NAFLD over the past decades, so that, in most countries, NAFLD has become the most frequent cause of liver disease.4–7 There has been a perception that some Asians who have NAFLD are not obese,5 but this may depend partly on the problems of ethnic-specific definitions of overweight and obesity.8,9 Conversely, not all obese persons have NAFLD.10,11 What is becoming apparent is that not all body fat is evil; some deposits are metabolically unhealthy, others are not!12–14 This issue of the Journal features four articles that have a bearing on factors linking obesity to NAFLD.15–18 The findings have implications for diagnosis and for targeting management strategies.2–5,18 To give context to these articles, this editorial focuses on factors that determine the risk of liver disease in people who are overweight or obese.

Relative bodyweight

There is a strong curvilinear relationship between body mass index (BMI) and relative body fat mass. For this reason, BMI is widely used to define overweight and obesity, although appropriate adjustments must be considered for gender and ethnicity.8,9 Contemporary clinical and epidemiological studies from China, Japan and Korea indicate that fatty liver is more often associated with obesity than with alcoholism (although the latter remains important!).5–7,15,19 There is also a direct association between BMI, extent of hepatic steatosis, non-alcoholic steatohepatitis (NASH), and advanced liver fibrosis.2,3,5 Overall, approximately 75% of obese subjects have steatosis, approximately 20% have NASH, and only approximately 2% have cirrhosis.1 In a recent editorial in Journal of Gastroenterology and Hepatology and elsewhere, we have discussed the invariable relationship between NAFLD and insulin resistance, and the close nexus between fatty liver and metabolic syndrome.20,21 It is now important for hepatologists to understand what it is about some cases of obesity that leads to insulin resistance and hepatic steatosis.

One factor that modifies NAFLD risk is recent weight gain.5 Weight gain of 5 kg or more since the ages of 18 to 20 years in both sexes also increases the risk of developing diabetes, hypertension, and coronary heart disease, and the risk of these conditions increases with the amount of weight gained.1 In light of the close ties between NAFLD and metabolic syndrome,20 weight gain would be expected to increase the risk of fatty liver disease; this occurs, even among some apparently ‘lean’ individuals.5 Conversely, even a very modest reduction in BMI (less than 5% of initial weight in the past 6 months) can reduce the size of selected fat depots, as reflected by improved liver function tests in obese patients.2–6,22 With moderate weight loss, whether achieved by ‘lifestyle intervention’ (the combination of increased physical activity and dietary adjustments) or by bariatric surgery, improvement in NAFLD is impressive.6,21,23,24 Improved aerobic fitness also decreases risk of developing obesity-associated complications.1 Of interest to the present discussion, a higher level of habitual physical activity is associated with lower intra-hepatic fat content.25

Body fat distribution

It should be emphasized that the distribution of adiposity (fat storage tissue) may be more important to hepatic fat accumulation (steatosis) than the total adipose mass.5 Obese persons with excess visceral adipose tissue (VAT, or abdominal obesity) are at higher risk for metabolic syndrome components than those whose fat is located predominantly in the lower body, subcutaneously.1,14 Further, ‘lean’ NASH patients usually have abdominal obesity, or more VAT.1,5 Waist circumference is highly correlated with abdominal fat mass (VAT) and, therefore, should now be used as a surrogate marker for abdominal obesity; such has been recommended by the Asia–Pacific Guidelines on NAFLD.4 A recent elegant study has shown that ob/ob mice, which are genetically predisposed to obesity, type 2 diabetes, metabolic syndrome and fatty liver, can be rescued from all these metabolic complications by expression of an adiponectin transgene.13 The biological effect of such forced expression of adiponectin was to expand the subcutaneous adipose tissue (SAT) mass.13 The observation that increasing adiposity in this adiponectin-transgenic ob/ob mouse protects against fatty liver and metabolic syndrome when fat distributes into its physiological storage sites (under the skin!) is consistent with the proposal that fat only distributes into the liver and other non-physiological sites (skeletal and cardiac myocytes, pancreatic beta-cells) when access to SAT becomes restricted –  that is, when it is full up!1,13 This conclusion is given further support by the observation that persons with morbid obesity and a high ratio of central adiposity (VAT) to total body fat have an increased risk of fatty liver and metabolic syndrome than those with lower ratios.11

Clinical and epidemiological studies using waist circumference to estimate VAT mass find a direct association between abdominal fat and liver fat content, independent of total adipose mass and BMI.5,11,26,27 Further, several recent studies have adopted imaging methods to demonstrate that VAT rather than subcutaneous abdominal fat (i.e. SAT) is more influential than BMI in terms of predicting the presence of fatty liver.11,26–30 In one such study, obese subjects with hepatic steatosis had higher ratios of truncal-to-total body fat compared with similarly obese persons without fatty liver.28 However, the sensitivity and specificity of available imaging methods for measuring liver fat content are unsatisfactory compared to [1H]-magnetic resonance spectroscopy; and the latter modality is usually not available outside a research setting.11,25,28–31 Conflicting results on the accuracy of different modalities have been reported, which may be related both to the inadequacy of an assay ‘gold standard’ (liver biopsy is rarely performed in this experimental context), and to technical factors.30,31

The first article on liver fat in this issue of the Journal comes from individuals participating in the Framingham Heart Cohort Study.18 The authors showed that one cross-sectional computed tomographic (CT) slice taken through the abdomen, but not through the chest, is adequate to capture most of the variance of liver fat content. Further, three measures in liver and two in spleen were as good as a larger number of slices for optimal precision in hepatic fat estimation. These findings of a reliable, simplified imaging method for hepatic fat determination based on limited CT slices should be helpful in future research studies, of the type described next.

Few studies have been performed to establish correlations between biopsy-proven hepatic steatosis and markers of obesity in a disease-free population. Also, in this issue of the Journal, Park et al. report a retrospective analysis of 177 living liver donors without a history of alcohol abuse.15 They evaluated the relationship between histological hepatic steatosis and clinical/anthropometric characteristics, including abdominal fat distribution measured by CT in living liver donors. By unpaired t-test, age, serum lipid profiles, BMI, liver/spleen Hounsfield ratio, VAT and SAT area were each associated with histological steatosis. However, in the multiple logistic regression analysis, only VAT and serum triglyceride concentration remained as independent risk factors for steatosis. Subgroup analysis confirmed that VAT was an independent risk factor for steatosis in both men and women.15 These findings strengthen the conclusion that VAT is more important than BMI or overall obesity for linkage to hepatic steatosis, even in apparently healthy liver donors.

Ethnicity and gender

Obesity-associated health risks are also influenced by ethnicity and gender. Asians generally have a higher percentage of body fat and VAT than Caucasians of the same age, sex and BMI.1,5,8,9 It therefore seems logical that lower cut-off values for BMI and waist circumference would be appropriate for Asian populations. This has been recommended by the International Diabetes Federation,9 but not by a WHO Expert Committee.8 A practical definition of obesity would ideally be based on the relationship between BMI and health outcome, rather than simply on body composition.1,4,5,9 In this respect, there appears to be a higher percentage of patients with NAFLD who are non-obese among Asians compared to their European counterparts.5

The impact of gender on BMI-related fatty liver is partly due to the difference in relative fat content, but also depends on the distribution of body fat.5 Women typically have a lower body distribution of adipose (gyneic pattern), characterized by peripheral subcutaneous fat deposition (SAT) in the gluteofemoral region. The latter means that the ratio of SAT to muscle area in the thigh is higher in women than men. Men have a different pattern of adipose distribution (android pattern), characterized by a higher proportion of fat in intra-abdominal regions (VAT), and lower ratio of SAT to thigh muscle area.1,14 Such relationships are explored further by Jun et al. in this issue of the Journal of Gastroenterology and Hepatology.16

The authors used CT to study the association between sonographic fatty liver and various distribution patterns of adipose tissue in 408 non-drinkers who were overweight or obese.16 The mean age was 47 years, 176 subjects were male, and the mean BMI was 26 kg/m2. The results show clearly different adipose distribution between the genders, independent of BMI. In females, the abdominal and thigh SAT areas were more extensive than in males, whereas the proportions of intra-abdominal fat area (VAT) and thigh muscle area was more extensive in males than in females. Even among subjects with normal BMI, the frequency of NAFLD increased with visceral fat area (VAT). It is also of interest that for women, but not for men, the frequency of NAFLD decreased for each quartile increase of thigh SAT fat, independent of BMI, age and VAT. This negative correlation between more extensive thigh SAT and NAFLD risk in women was confirmed by logistic regression analysis. These observations support the concept that women tend to have a larger area of femoral subcutaneous adipose and a smaller area of VAT than men.1,16 Thus, the capacity (or restrictiveness) of peripheral fat deposits may influence the proclivity to visceral adiposity and, thereby, exert an opposite effect on the development of fatty liver and other metabolic complications (protects vs facilitates).31

Genetic polymorphisms

As already mentioned, more than two-thirds of obese individuals develop hepatic steatosis, but only a minority develops NASH and cirrhosis. This raises the question about which factors determine whether a patient is ‘at risk’ for developing more complicated forms of NAFLD. The distribution of lipid stores in the body discussed above may be important, because free fatty acids (FFA) arising from active lipolysis of the portal venous drainage area adipocytes (VAT) are taken up by the liver to become the major source of lipids contributing to hepatic triglyceride stores in NAFLD.2,11,21,32,33 The functional differences between the visceral and the subcutaneous adipocytes may be related to their anatomical location. VAT and its adipose-tissue-resident macrophages produce more pro-inflammatory cytokines like tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6) and less adiponectin. These cytokine changes induce or compound insulin resistance,21 and play a major role in the pathogenesis of hepatic steatosis and subsequent fibrosis.2,3,11,14 An attractive hypothesis is that polymorphisms in genes that influence the release of adiponectin, or elaboration of TNF-α and IL-6, contribute to phenotypic differences in NAFLD, and particularly why some cases develop NASH complicated by hepatic fibrosis.2,3,11,13,21,33

Also in this issue of the Journal, Wong et al. report their studies into genetic polymorphisms of adiponectin and TNF-α in 79 Chinese patients with histologically proven NAFLD/NASH.17 Together with 40 controls, patients were tested for nucleotide polymorphisms of adiponectin sequence −11 391, −11 377, +45 and +276, and of the TNF-α promoter at −863, −308 and −238. The results failed to identify adiponectin and TNF-α gene polymorphisms in association with NAFLD, or with significant fibrosis. The adiponectin −11 377G and +45G alleles were associated with hypertriglyceridemia, but not with disease severity. Unfortunately, like most similar investigations in this field, this small study was not adequately powered to detect subtle differences in allele frequencies.3,17 It is hoped that regional cooperation in Asia will be used to generate larger-sized studies across a diversity of different ethnic groups to establish or refute the influence of gene polymorphisms as susceptibility factors for progressive NASH.

Conclusions and future directions

In summary, although obesity is closely associated with NAFLD, the distribution of excess bodily fat storage may be just as or even more important. In particular, VAT is the most important factor for the development of hepatic steatosis, independent of total adipose mass, whereas subcutaneous fat storage capacity is protective.1,5,14,32,33 Gender, age and genetic factors are likely to influence bodily distribution of fat. As already elegantly demonstrated in transgenic mice,13 and humans treated with ‘glitazones’, peroxisome proliferator-activated receptor (PPAR)γ agonists which promote differentiation of pre-adipocytes to expand SAT, the redistribution of fat from visceral to peripheral subcutaneous storage sites has the potential to correct fatty liver and metabolic syndrome.32–35 These findings also support the portal hypothesis, where FFA and other factors released from VAT contribute to increased hepatic lipid stores, to inflammatory recruitment and, likely, to hepatic and peripheral insulin resistance.2,3,11,21 More carefully conducted longitudinal genetic studies are needed to obtain additional information on how and why variations occur in different lipid stores (VAT and SAT), and the relationship of these changes to the development of NAFLD/NASH, diabetes and metabolic syndrome. It is already known that high VAT and high liver fat content are associated with resistance to lifestyle intervention.29 It therefore seems evident that efforts to prevent or correct NAFLD should be directed at reducing VAT stores during the earliest stages of their development, as indicated by recent weight gain and detected by increased waist circumference.4,5,23 The latter anthropometric finding can be backed up by judicious imaging (e.g. by CT with limited abdominal slices), as confirmed by the findings of Speliotes and colleagues.18 Early intervention has the greatest potential to decrease liver fat and reduce the risk of obesity-related metabolic problems.1,4,5,23 Future studies are required to determine how much visceral fat reduction is needed to induce favorable metabolic changes in the liver.