Dr Z. M. Younossi, Center for Liver Diseases at Inova Fairfax Hospital, 3289 Woodburn Road, Suite 375, Annanadale, Virginia 22003-6800, USA. E-mail: firstname.lastname@example.org
Background Several adipocytokines have been implicated in the pathogenesis non-alcoholic fatty liver disease (NAFLD).
Aim To assess adipocytokines in NAFLD patients and controls.
Methods A total of 95 patients (26 non-alcoholic steatohepatitis (NASH), 19 simple steatosis (SS), 38 obese controls and 12 non-obese controls) were included. Fasting serum insulin, glucose, visfatin, resistin, adiponectin, tumour necrosis factor-α (TNF-α), interleukin-8 (IL-8) and IL-6 were determined. Univariate and multivariate analyses were used to compare groups and determine associations.
Results Serum TNF-α and IL-8 were higher in NAFLD patients when compared with both obese and non-obese controls. Analysis involving all patients revealed a significant correlation between serum TNF-α and IL-8 (P < 6.319e−08), and between IL-6 and IL-8 (P < 5.271e−15). Homeostatic model assessment scores negatively correlated with adiponectin in NAFLD (P < 0.0032). Serum visfatin was higher in all three obese groups than in non-obese controls (P < 0.02, P < 0.002 and P < 0.008). Visfatin in NASH patients was lower than SS and obese controls. Although TNF-α was associated with NAFLD (P < 0.02), it was interdependent on visfatin. In comparison to SS, four factors were independently associated with NASH: age, alanine aminotransferase, IL-8 and adiponectin (P < 0.05). Multivariate analysis indicated that TNF-α was the only independent predictor of fibrosis in NASH (P < 0.0004).
Conclusion These findings support a complex interaction between adipocytokines and the pathogenesis of NAFLD.
Non-alcoholic fatty liver disease (NAFLD) is the hepatic manifestation of metabolic syndrome.1, 2 The impact of NAFLD relates to its prevalence and potential for progression. NAFLD is the most common cause of liver test abnormalities among adults,3–5 with a prevalence of 10–24% in the general population. The estimated prevalence of non-alcoholic steatohepatitis (NASH) is between 3% and 5%.3, 6 However, NAFLD occurs in 75–95% of some patient populations, such as the morbidly obese; the estimated prevalence of NASH in this population is between 20% and 47%.3, 6, 7 Given the increasing prevalence of obesity, NAFLD incidence is on the rise.
Non-alcoholic fatty liver disease is characterized mainly by excessive hepatic deposition of free fatty acids and triglycerides in the hepatic parenchyma.6, 8 The pathological spectrum of NAFLD ranges from simple steatosis (SS) to NASH, potentially leading to fibrosis and cirrhosis.9 Important differences in the management strategies for SS and NASH make the distinction between these two types of NAFLD critical.4, 9 Currently, the only definitive diagnostic tool for the progressive forms of NAFLD is liver biopsy.10, 11
The most widely accepted explanation for NASH pathogenesis is the multiple hit hypothesis.12, 13 According to this hypothesis, the first hit is an increase in the fatty acids synthesis and the fat uptake by liver cells.14 This process is indirectly triggered by insulin resistance and leads to macrovesicular steatosis.13, 14 The second ‘hit’ may involve increases in fatty acid beta oxidation, oxidative stress, pro-inflammatory cytokines and endotoxemia.13, 14 One of these pathways may be the predominant ‘second hit’ in a specific cohort of patients with NASH. Increasing evidence indicates that white adipose tissue (WAT)-derived adipokines also contribute to the pathogenesis of NASH.13–17 Among these adipokines, adiponectin seems to play a prominent role in the pathogenesis of NASH. In fact, hypoadiponectinemia was recently shown to be responsible for the accumulation of hepatic fat as well as the development of the insulin resistance.15, 20, 21 On the other hand, the role of resistin (an adipokine associated with inflammatory conditions such as insulin resistance, obesity and atherosclerosis) and visfatin (an adipokine that exerts both insulin-mimetic and immunomodulating effects) in the pathogenesis of NASH is less clear. Another layer of complexity in the pathogenesis of NAFLD involves an interplay between adipokines and pro-inflammatory cytokines produced both by the peripheral blood mononuclear cells and infiltrating lymphocytes, as well as macrophages embedded in white adipose tissue.18
This study profiles several cytokines and adipokines in patients with NASH, SS and controls, and assesses their interactions with each other and underlying histopathological changes.
Materials and methods
The study subject included 95 patients, 45 of whom had NAFLD (NASH, n = 26 and SS, n = 19): 38 patients were designated as obese controls without NAFLD (control I, n = 38); and 12 patients were designated as non-obese Controls who were healthy blood donors without evidence of liver disease (control II, n = 12). Patients with other liver diseases, excessive alcohol use (>10 g/day) and treatment with PPARγ agonists were excluded.
Each liver biopsy specimen was fixed in formalin, routinely processed for histology, sectioned and stained with haematoxylin–eosin and Masson trichrome. The degree of steatosis was assessed in haematoxylin–eosin-stained sections and graded as an estimate of the percentage of tissue occupied by fat vacuoles as follows: 0 = none; 1 = <5%; 2 = 6–33%; 3 = 34–66%; 4 = >66%. Other histological features evaluated in haematoxylin–eosin sections included portal inflammation, lymphoplasmacytic lobular inflammation, polymorphonuclear lobular inflammation, Kupffer cell hypertrophy, apoptotic bodies, focal parenchymal necrosis, glycogen nuclei, hepatocellular ballooning and Mallory bodies. These histological features were graded as follows: 0 = none; 1 = mild or few; 2 = moderate and 3 = marked or many. Fibrosis was assessed with the Masson trichrome stain. Portal fibrosis and interlobular pericellular fibrosis were graded as follows: 0 = none; 1 = mild; 2 = moderate; and 3 = marked. When present, bridging fibrosis was noted as few or many bridges, and cirrhosis was identified by parenchymal nodules surrounded by fibrous tissue. Cirrhosis was further categorized as incomplete or established, depending on the degree of loss of acinar architecture. Each liver biopsy was assigned to one of four diagnostic categories: (i) no fatty liver disease present (obese controls), (ii) bland or simple steatosis (SS), (iii) steatosis with nonspecific inflammation or (iv) NASH. Patients were defined as having bland or simple steatosis (SS) if they had any degree of hepatocellular fat accumulation as their sole pathology. Pathologic criteria for NASH included steatosis, ballooning degeneration and lobular inflammation or at least one unequivocal Mallory body on the haematoxylin and eosin stain, or some degree of zone 3 pericellular fibrosis or bridging fibrosis identified on the trichrome stain. The obese control group included patients whose liver biopsies showed no evidence of fatty liver disease.11
Serum samples were collected at the time of liver biopsy, immediately frozen in −80 °C, and subsequently used for biochemical analyses such as glucose and insulin measurements and for adipocytokines profiling.
Groups 1–3 (NASH, SS and Obese controls) were matched for body mass index (BMI) and insulin resistance as measured by homeostatic model assessment (HOMA) scores (Table 1). Furthermore, each of these three groups was subdivided according to HOMA scores for the subsequent analyses: high HOMA (>3); low HOMA (<2); and mid-range HOMA (2–3).
Table 1. Clinico-demographic and laboratory data (mean ± s.d. or %)
Liver histology was available for non-alcoholic fatty liver disease patients as well as obese controls.
This study was reviewed and approved by the Institutional Review Boards of Inova Fairfax Hospital and of George Mason University.
Glucose levels were measured by Glucose Oxidase-based kits (Sigma-Aldrich, St Louis, MO, USA) according to manufacturer’s protocol. Insulin levels in serum samples were quantified by sandwich ELISA (LINCO Research, St Charles, MO, USA). HOMA scores were obtained with homa calculator software, version 2.2 (http://www.dtu.ox.ac.uk/).
Measurements of adipokines and cytokines
Serum levels of adipokines and cytokines were measured with enzyme immunoassays according to the manufacturers’ instructions. Each measurement was performed in duplicate. Tumour necrosis factor-α (TNF-α), interleukin-6 (IL-6) and IL-8 were quantified by using Compact ELISA kits from RDI Division of Fitzgerald Industries Intl (Concord, MA, USA). Resistin levels were assessed with kits provided by BioVendor Laboratory Medicine, Inc (Candler, NC, USA). Adiponectin and visfatin serum levels were measured with competitive ELISA assays from Phoenix Pharmaceuticals, Inc. (Belmont, CA, USA) according to manufacturer’s manual. The calibration curves were constructed by plotting the net average absorbances of the standards on the Y-axis and the concentrations on the X-axis using the logit-log function to linearize and draw the best fitting curve. Concentrations of the adipocytokines in each sample were calculated from the calibration curve with sigma plot software v. 7. The correlation coefficients were linear in a concentration range between 1 and 700 pg/mL for TNF-α (r = 0.973); 2 and 300 pg/mL for IL-8 (r = 0.989); 1.5–400 pg/mL for IL-6 (r = 0.968); 1.82–49 ng/mL for visfatin (r = 0.977); 1.5–50 ng/mL for resistin (r = 0.969); 0.3–100 μg/mL for adiponectin (r = 0.975); and between 2 and 200 μU/mL for insulin (r = 0.982). The samples with higher concentrations of analytes were quantified after dilution.
Pairwise comparisons of the serum concentrations of insulin, glucose, visfatin, TNF-α, resistin, adiponectin, IL-8 and IL-6 were performed between groups of patients with the nonparametric Mann–Whitney test. The one-way anova was used to compare three or more groups. Associations between the concentrations for pairs of adipokines and cytokines of interest were tested with use of Pearson correlation coefficients after appropriate log-normalizations of concentration values. Additionally, multivariate linear regressions with stepwise variable selection were used to test for significant relations in continuous data with adjustment for possible confounders.19 Categorical data were subjected to univariate and multivariate ordered probit regression analysis. Unless otherwise noted, we used two-tailed hypothesis tests and P-values < 0.05 were considered significant.
Clinical and demographic data are summarized in Table 1. Non-obese controls were healthy blood donors, 50% of whom were females. Their average BMI was 24.9 ± 1.8 and their average age was 56 ± 17 years with normal aminotransferases and no evidence of viral hepatitis and other chronic liver diseases. Comparative measurements for fasting serum levels of insulin, glucose, visfatin, resistin, adiponectin, TNF-α, IL-6 and IL-8 levels were performed in all NAFLD patients and controls (see Supplementary Table S1). The comparisons are as follows.
Comparison of adiponectin, resistin and visfatin in different groups
Neither groupwise comparisons nor analysis of correlations revealed statistically significant findings related to serum resistin levels in NASH, SS and the obese controls (Control I). Serum visfatin levels were significantly higher in the obese controls compared with healthy, non-obese controls. Interestingly, visfatin levels in NASH patients were lower in comparison with patients with SS and the obese controls (Figure 1a).
Homeostatic model assessment scores negatively correlated with serum adiponectin levels in all NAFLD patients (r = −0.44, P < 0.0032), but not in the control I cohort. Groupwise comparisons showed that NASH patients have a significantly (P < 0.001) lower adiponectin level than those with SS and the obese controls without NAFLD (NASH 6.7 ± 6.5 μg/mL, SS 12.2 ± 7.56 μg/mL, obese control or Control I 10.2 ± 7.4 μg/mL). On the other hand, stratification of patients based on HOMA scores (high-HOMA group vs. low-HOMA group) failed to show a significant difference in adiponectin levels among different groups.
Comparisons of pro-inflammatory cytokines in different groups
Serum levels of TNF-α and IL-8 were significantly higher in NAFLD patients compared with both obese and non-obese controls. Additionally, obese controls showed lower serum levels of TNF-α than non-obese controls (Figure 1b,c). IL-6 profiling revealed statistically significant differences between NAFLD and obese controls, between NASH and obese controls, and between SS patients and non-obese controls (Figure 1d). When patients with SS were compared to the obese controls and to patients with NASH, the differences failed to reach statistical significance (Figure 1d).
When all the patients profiled in this study were stratified according to their HOMA scores (high-HOMA group vs. low-HOMA group), the differences in TNF-α levels between these groups remained significant (P < 0.01). Similar observations were made for IL-8 and IL-6 levels (both P-values < 0.01).
A correlation analysis was evaluated the strength of the association between clinical parameters and serum adipocytokines. This analysis revealed a significant correlations between serum concentrations of TNF-α and IL-8 (R = 0.5276, P < 6.319e−08), as well as between IL-6 and IL-8 (R = 0.7079, P < = 5.271e−15). On the other hand, TNF-α and IL-6 levels showed no correlation. Interestingly, the association of TNF-α and IL-6 in patients with SS reached statistical significance (R = 0.4583, P < 0.05). Similarly, this correlation remained significant when obese controls and SS patients were profiled together (R = 0.4689, P < 0.0003).
Additionally, the multivariate analysis indicated that in NAFLD patients, IL-6 levels depend on HOMA scores (P < 0.032) and IL-8 levels (P < 0.0027) (Table 2). Furthermore, in NAFLD patients, TNF-α levels were dependent on serum glucose concentration, HOMA, BMI and IL-8 (P < 0.01).
Table 2. Best fitting multiple linear regression models showing the relationship between IL-6 and other parameters. Regression coefficient β represents slope estimate ± standard error of the estimate (S.E). Positive and negative regression coefficients indicate the dependence of IL-6 levels on levels of other soluble serum components in the particular group of patients
Note that when all obese patients were subjected to multiple regression analysis, the concentrations of TNF-α were significantly dependent on resistin levels (P < 0.0176). On the other hand, in obese controls and in NAFLD patients, the correlations of TNF-α and resistin serum levels were insignificant but followed opposite trends (Table 3).
Table 3. Best fitting multiple linear regression models showing the relationship between TNF-α and other parameters. Regression coefficient β represents slope estimate ± S.E. of the estimate (S.E.). Positive and negative regression coefficients indicate the dependence of TNF-α levels on levels of other soluble serum components in the particular group of patients
TNF-α, tum our necrosis factor-α; HOMA, homeostatic model assessment; IL-8, interleukin-8; NAFLD, non-alcoholic fatty liver disease; BMI, body mass index.
Regression coefficient β and S.E.
P-values of independent variables
P-value of the entire model
−3.0775 ± 1.5765
1.2510 ± 0.6087
1.3436 ± 0.6010
−1.3200 ± 0.6019
0.1027 ± 0.0446
−0.0690 ± 0.0391
0.4776 ± 0.9365
−1.0073 ± 0.3303
−0.6064 ± 0.2575
0.7190 ± 0.2315
1.2745 ± 0.4593
0.1531 ± 0.0885
0.4399 ± 0.0869
All obese patients Profiled
0.2119 ± 0.0660
0.3343 ± 0.0421
−0.1629 ± 0.0619
Adipokines and cytokines levels differentiate NAFLD from obese controls
The serum levels of all the profiled cytokines (TNF-α, IL-6 and IL-8) were significantly different in NAFLD patients and the obese controls. On the other hand, visfatin, resistin and adiponectin levels were not different between these two groups (Table 4). The levels of all three cytokines were higher in NAFLD than the obese controls. Observed differences remained significant after multiple test adjustment. After taking into account all additional clinical parameters listed in Table 1, no single noncytokine parameter seemed to be independently associated with NAFLD. According to the multivariate analysis, TNF-α positively correlated with NAFLD (P < 0.02), but was interdependent with the levels of the visfatin.
Table 4. Groupwise comparisons of serum adipokine and pro-inflammatory cytokine levels for patients with NAFLD and obese controls
Adipokines and cytokines levels differentiate NASH from simple steatosis
Serum levels of adipocytokines in patients with NASH and Simple Steatosis revealed some interesting differences (Table 5). TNF-α levels were higher in NASH patients, whereas levels of IL-8, visfatin and adiponectin were lower in NASH patients. After multiple test adjustments, only the differences in TNF-α and adiponectin remained significant (both with P-values < 0.003). When additional clinical parameters (Table 1) were taken into account in a multivariate analysis, the results revealed that age and alanine aminotransferase (ALT) positively correlated with histologic NASH (P-values < 0.01 and < 0.02, respectively), whereas IL-8 and adiponectin negatively correlated with histologic NASH (P-values < 0.05 and <0.03, respectively). On the other hand, both TNF-α and HOMA-IR failed to be independently associated with histologic NASH.
Table 5. Groupwise comparisons of serum adipokine and pro-inflammatory cytokine levels for patients with simple steatosis and NASH
Adipokine and cytokine levels differentiate different stages of hepatic fibrosis
In addition to the subtypes of NAFLD, we also assessed the relationships between hepatic fibrosis and serum adipokines and cytokines. A multivariate regression analysis determined whether any of the adipokine or cytokine predicted the stage of fibrosis. This analysis indicated that only serum TNF-α was an independent predictor of histological fibrosis in patients with NASH (P < 0.0004). No other variable tested independently or in combination behaved as reliable predictor of the stage of fibrosis.
This study provides an in-depth analysis of a large number of cytokines and adipokines in well-characterized cohorts of patients with NAFLD. The rationale for this study was to cover the gap in the existing literature describing the relationship between several adipokines and cytokines potentially implicated in the development of NAFLD and its progressive form (NASH). The three important adipokines (adiponectin, resistin and visfatin) and three important pro-inflammatory cytokines (TNF-α, IL-6 and IL-8) were profiled in well-matched cohorts of NAFLD patients and controls for whom extensive clinical and pathological data were available. This study confirms the previously reported findings that hypoadiponectemia is negatively associated with HOMA.20, 21 Additionally, lower levels of adiponectin are associated with histologic NASH. Finally, we confirmed the absence of a significant association between NAFLD subtypes and serum resistin levels.15
Our study provides some new insights into adipocyokines in NAFLD. Our data show that TNF-α levels increased significantly from obese controls, SS, to NASH, emphasizing the potential role of TNF-α in NAFLD (Tables 4 and 5). Earlier researchers had hypothesized that TNF-α plays a role in the pathogenesis of the diseases related to IR, including NAFLD and NASH.22 Indeed, elevated TNF-α production has been observed in cultures of peripheral blood cells collected from obese patients with NAFLD.23 Nevertheless, direct evidence of TNF-α involvement in the earlier stages of NAFLD was not previously described. The pathogenic role of TNF-α may be to attract inflammatory leucocytes to the liver, or increase SREBP-1c (sterol regulatory element binding protein-1c) dependent on intrahepatic fat deposition.24 Our observations corroborate findings by Satapathy et al.25 who treated patients with histologically proven NASH with a TNF-α inhibitor: pentoxifylline. Pentoxifylline treatment resulted in a significant reduction of serum TNF-α levels accompanied by normalization of both ALT and aspartate aminotransferase (AST), and improvements in the insulin resistance index. Together, these studies support the potential role of anti-TNF-α in the treatment of NASH.
The role of IL-6 in the pathogenesis of NASH is less clear. A previous study showed that both IL-6 and sIL-6R levels were increased in NASH patients compared to patients with simple steatosis and healthy volunteers.26 In our study, the serum levels of IL-6 were higher in obese controls and in patients with SS. In contrast, IL-6 levels were significantly lower in NASH patients in comparison with SS patients (Figure 1). This pattern can be explained by the failure of IL-6 and/or visfatin-dependent hepatoprotection associated with the onset of overt NASH. The protective function of IL-6 in steatotic livers is related to its ability to suppress oxidative stress and mitochondrial dysfunction,27, 28 and prevent the release of reactive oxygen species, mitochondrial permeability transition and ethanol-mediated depletion of adenosine triphosphate.28 IL-6 also provides hepatoprotection in ischaemic preconditioning models29 and prevents obesity and alcohol-associated fatty liver and its related dysfunction in transplanted livers.27, 30 In our cohort of NASH patients, IL-6 levels were independent of other variables potentially participating in the development of NASH (data not shown). On the other hand, in the entire NAFLD cohort, IL-6 levels depended on HOMA scores (P < 0.032) and IL-8 (P < 0.0027) levels but not on TNF-α levels.
Our study suggests that visfatin levels in various NAFLD groups resemble IL-6 levels, suggesting that they may be co-regulated. Levels of IL-6 and visfatin might be correlated because of the presence of both negative and positive feedback loops. For example, visfatin induces the production of IL-6 in human CD14 (+) monocytes,28 whereas IL-6 negatively regulates visfatin gene expression in 3T3-L1 adipocytes.29 Serum visfatin levels were significantly higher in each of the subgroup of obese patients when these groups were compared with non-obese controls (control II). On the other hand, visfatin levels were lower in NASH patients than in obese controls (control I) and in patients with SS. To our knowledge, this is the first clinical observation pointing to the potential involvement of visfatin in the pathogenesis of NAFLD.28–32
Our analysis also focused on the use of adipocytokines to differentiate NAFLD from appropriately matched controls. Our data suggest that TNF-α is important in the pathogenesis of NAFLD. On the other hand, in comparison to SS, NASH is independently associated with age, ALT, adiponectin and IL-8. These important observations suggest a role for TNF-α in promoting NAFLD as well as the role of hypoadiponectinemia and IL-8 in the development of NASH. Additionally, these findings indicate the potential for TNF-α, IL-8 and adiponectin in the development of non-invasive diagnostic biomarkers for NASH.
Finally, we analysed the association between hepatic fibrosis and serum adipokine–cytokine levels. This analysis revealed that TNF-α is the only factor associated with hepatic fibrosis in NASH. These findings are in agreement with observations collected on the mouse model of NASH, which is deficient in TNF receptors 1 and 2. Wild–type mice and those with deficient TNF receptors 1 and 2 were fed a methionine and choline deficient (MCD) diet.33 Despite the MCD diet, mice with deficient TNF receptors 1 and 2 showed impairments in both the activation of hepatic stellate cells and mRNA expression of tissue inhibitor of metalloproteinase 1, leading to a phenotype with a less fibrogenic liver.33 Further analysis of TNF-α-related fibrogenesis in liver specimens of patients with NASH is needed.
In conclusion, this study provides an in-depth assessment of several important adipokines and cytokines in a well described and well controlled group of patients with NAFLD. Our findings confirm that hypoadiponectinemia and IL-8 are consistently associated with NASH. TNF-α is implicated in the development of NAFLD and NASH-related fibrosis. Furthermore, visfatin and IL-6 may play a protective role in NAFLD. Together, these findings show the complexity of the interactions between various adipokines, IR and the pathogenesis of NAFLD.
Declaration of personal interests: None. Declaration of funding interests: This study has been supported in part by the Liver Disease Outcomes Fund of the Center for Liver Diseases at Inova Fairfax Hospital, Inova Health System and by internal funding from the College of Science at George Mason University.