SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Although nonalcoholic fatty liver disease (NAFLD) is frequent in obesity, the metabolic determinants of advanced liver disease remain unclear. Adipokines reflect inflammation and insulin resistance associated with obesity and may identify advanced NAFLD. At the time of obesity surgery, 142 consecutive patients underwent liver biopsy and had their preoperative demographic and clinical data obtained. Liver histology was scored by the NAFLD activity score, and patients subdivided into four groups. Concentrations of retinol-binding protein 4 (RBP4), adiponectin, tumor necrosis factor-α (TNF-α), and leptin were determined ∼1 week prior to surgery and results were related to liver histology. The prevalence of no NAFLD was 30%, simple steatosis 23%, borderline nonalcoholic steatohepatitis (NASH) 28%, and definitive NASH 18%. Type 2 diabetes mellitus (T2DM) and metabolic syndrome (MS) prevalence were 39 and 75%, respectively, and did not differ across the four histological groups (P = NS). Triglyceride (TG) and alanine transaminase (ALT) levels, strongly associated with advanced stages of NAFLD and NASH (P = 0.04). TG levels >150 mg/dl, increased the likelihood of NASH 3.4-fold, whereas high-density lipoprotein (HDL) levels predicted no NAFLD (P < 0.01). Concentrations of TNF-α, leptin, and RBP4 did not differ among histological groups and thus did not identify NASH; however, there was a trend for adiponectin to be lower in NASH vs. no NAFLD (P = 0.061). In summary, both TG and ALT levels assist in identification of NASH in an obesity surgery cohort. These findings underscore the importance of fatty acid delivery mechanisms to NASH development in severely obese individuals.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Nonalcoholic fatty liver disease (NAFLD) is extremely frequent in patients with obesity and type 2 diabetes mellitus (T2DM) and can develop into liver failure and cirrhosis (1,2,3). Hepatic steatosis, a consequence of hepatic insulin resistance and hyperinsulinemia (4), is closely associated with T2DM and metabolic syndrome (MS) (2,5,6). Hepatic steatosis plays a critical role in the development of MS/T2DM (7) and contributes to hyperglycemia, hyperlipidemia, and systemic subclinical inflammation present in these conditions (7,8,9,10). NAFLD represents a clinical spectrum from mild steatosis to inflammation and fibrosis, resulting in steatohepatitis (nonalcoholic steatohepatitis (NASH)) (3,11). However, the presence and progression of NAFLD is difficult to detect clinically and liver biopsy remains the gold standard for diagnosis and prognosis for NAFLD (1). Abnormal liver transaminase levels are nonspecific and thus insufficient for clinicians to determine which obese subjects have steatosis alone from those with advanced disease. Identifying advanced or progressive disease may help in selecting patients for liver biopsy and assist with more aggressive therapeutic options.

A number of studies have clearly associated more severe liver injury (NASH) with features of MS in clinically severe obesity (1,5,6). Fundamental components of this syndrome include insulin resistance, central obesity, dyslipidemia, and hypertension (8,12). Previous studies in obesity surgery cohorts have identified various features of MS, especially waist circumference, hyperglycemia, hypertension, and insulin resistance to associate with progression to fibrosis, regardless of severity of global adiposity (1,2,3). Various pro- and anti-inflammatory cytokines secreted from abdominal adipose tissue (10,13,14) have been identified to link to the development of comorbidities associated with obesity, particularly T2DM and MS (13,15). These secretory products, known as adipokines, have been shown to regulate key processes including insulin mediated glucose metabolism, fatty acid utilization, inflammation, and lipid overloading of ectopic tissues, such as liver (8,13,16).

Given our increasing understanding of the role of adipokines in modulating metabolic disease, it is unclear whether a comprehensive panel of specific adipokines with mechanistic links to pathophysiologic processes may serve as biomarkers for severity of NAFLD in obese patients. A growing body of evidence supports a role for the inflammatory cytokine, tumor necrosis factor-α (TNF-α), to mediate obesity associated insulin resistance, liver injury, and fibrinogenesis (13,16). In contrast, other adipokines such as adiponectin and leptin demonstrate antidiabetogenic and anti-inflammatory properties and thus may mediate protection from liver inflammatory processes and injury (13,16). Accordingly, a disproportionate increase in the TNF-α-to-adiponectin ratio has been proposed as a potential novel risk indicator for NAFLD progression to NASH (13,16). Although retinol-binding protein (RBP4) originates from the liver and is closely associated with adiposity, insulin resistance, and liver fat (17), no clear link has been established to advanced or chronic liver disease in severely obese subjects (18). Thus, the working hypothesis in this study is that adipokines closely linked to MS pathogenesis would relate to NAFLD severity, and consequently provide noninvasive tools to identify susceptible individuals in a cohort at high risk for NASH. The objective of this study was to determine the relationships between various clinical measures and adipokine measurements implicated in metabolic disease to histological categories of NAFLD, determined at the time of obesity surgery in a clinical cohort.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Subjects

One hundred and sixty patients were recruited through the Cleveland Clinic Foundation Bariatric and Metabolic Institute; 142 had a liver biopsy performed at the time of laparoscopic obesity surgery. Liver histology of patients was assessed by an experienced hepatopathologist (L.Y). The NAFLD activity score (19) was applied to each patient (described earlier). Demographic, clinical, and laboratory data were obtained from clinic visits 2 weeks prior to surgery. Blood was obtained for research repository concentration determination within 1 week of surgery. The samples from this repository were used to determine adipokine levels reported in this study. Presence of T2DM, hypertension, and hyperlipidemia was assessed by review of past medical history. Prevalence of T2DM was based on past medical history and/or fasting glucose of ≥126 mg/dl (20). BMI was computed as weight in kilograms (kg) divided by squared height in meters. Presence of MS was defined by the National Cholesterol Education Program Adult Treatment Panel III criteria of having three or more of the five components (2,12).

Preoperative assessment

Patients with a BMI >35 kg/m2, with significant medical, physical, or psychosocial disabilities meeting the National Institutes of Health criteria for obesity surgery were considered for entry into the program. Preoperative evaluation included visits with a multidisciplinary team consisting of nutritionist, psychologist, bariatric physician, and general surgeon. The baseline assessment included anthropometric measurements, history, and physician and laboratory tests. An assessment of alcohol consumption was obtained at the surgery consultation and separately by a psychologist and a self-reported questionnaire. Alcohol consumption in the cohort was required to be <20 g per week to be eligible for surgery. Laboratory tests included liver function tests, fasting glucose, glycosylated hemoglobin A1c, and lipid panel. Preoperative assessments were performed at the surgical consultation within 1 month of surgery date.

Liver biopsies were performed as a routine part of the operative procedure. The liver biopsies were taken from the left upper quadrant under laparoscopic view using a 14-gauge needle introduced percutaneously into the left lobe of the liver.

Liver histology

The histological diagnosis of NAFLD was established by the study pathologist (L.Y.). The NAFLD activity score was applied as described (19). Patients were subdivided into four histological groups: not NAFLD (normal liver biopsy), simple steatosis (steatosis in the absence of inflammation and ballooning hepatocyte degeneration), borderline NASH, (steatosis with minimal, rare inflammation, and hepatocyte ballooning), and NASH (steatosis with inflammation and hepatocyte ballooning, often with fibrosis). In brief, the degree of steatosis, liver injury, and inflammatory activity were scored using an 8-point scale (steatosis 0–3; lobular inflammation 0–3; ballooning hepatocyte degeneration 0–2). The NAFLD activity score is the unweighted sum of steatosis, lobular inflammation, and hepatocellular ballooning scores. The stage of fibrosis was scored using a 6-point scale (1a, b = mild (1a)/moderate (1b) zone 3 perisinusoidal fibrosis; 1c = portal fibrosis only; 2 = zone 3 and portal/periportal fibrosis; 3 = bridging fibrosis; 4 = cirrhosis). Patients with other forms of liver disease identified histologically were excluded from the study.

Adipokine assays

All adipokine assays were performed by a single experienced technician in the Endocrinology Core Laboratory of Cleveland Clinic. Leptin, high-sensitivity TNF-α, and total adiponectin concentrations were measured by sandwich enzyme-linked immunosorbent assay techniques (R&D Systems, Minneapolis, MN). The interassay coefficients of variation (mean ± s.d.) at three levels were of 9.7% (8.3 ± 7.8), 8.0% (231 ± 18.6), and 5.9% (435 ± 26) for leptin; 10.0% (2.3 ± 0.23), 6.4% (10.5 ± 0.67), and 5.2% (18.5 ± 0.96) for TNF-α; and were 8.3% (28.8 ± 2.4), 11.0% (87.5 ± 9.6), and 9.9% (163 ± 16) for total adiponectin, respectively. Serum RBP was also determined by sandwich enzyme-linked immunosorbent assay (ALPCO Diagnostics, Salem, NH) with interassay variations at two levels of 7.7% (1.2 ± 0.097) and 10.8% (2.4 ± 0.28).

Statistical analysis

Because of the skewed distributions of TNF-α, adiponectin, leptin, and the ratio of TNF-α-to-adiponectin, these variables were log10-transformed prior to analysis. We present the untransformed values within tables, but statistical analyses were performed on the log-transformed values. χ2-Tests on the binary variables MS and diabetes diagnosis were used to examine associations with liver histology. Triglyceride (TG) was dichotomized according to the threshold level of MS diagnosis. Mean and variance summary statistics are provided by liver histology group.

ANOVA was used to compare mean levels of biomarkers across the four liver histology groups; these biomarkers included RBP4, BMI, high-density lipoprotein (HDL), blood glucose, mean systolic blood pressure, aspartate aminotransferase, alanine transaminase (ALT), log(TG), log(leptin), log(adiponectin), log(TNF-α), and log(TNF-α/adiponectin). For these analyses, the liver histology groups were coded in three separate ways. First, they were analyzed as the four distinct diagnostic categories. Second, the groups, borderline NASH, simple steatosis, and NASH, were combined into one category and studied vs. no NAFLD subjects; this analysis contrasted diseased and nondiseased patients. Last, the analysis was implemented combining no NAFLD, borderline NASH, and simple steatosis groups to then examine these subjects against those with NASH; this analysis contrasted the most severe group, NASH, to the other three categories. Each of these three analyses tested a different hypothesis, therefore, when adjusting for multiple tests, only four adjustments were necessary. Pearson's correlation coefficients were calculated for all quantitative variables to assess the strength of associations between any two variables. Stepwise logistic regression was used to model how the categorical variable of TG, as described above, HDL, ALT, and leptin independently predicted liver histology. Two separate prediction models were constructed, the first focusing on NASH vs. all other histological groups, and the second focusing on the “healthy state” (no NAFLD) vs. all other histological groups.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Baseline characteristics

One hundred and forty two liver biopsies were performed at laparoscopic bariatric surgery from 2005 to 2007. The sample was 79% female (N = 112) and 21% male (N = 30). The ethnic distribution was 18% African American (N = 25), 81% white (N = 116), and <1% Hispanic (N = 1). The mean (±s.d.) age for the cohort was 49 ± 10 years and the mean BMI was 48 ± 7 kg/m2. In this group of 142 subjects, there were 107 subjects (75%) with National Cholesterol Education Program MS, 55 with T2DM (39%), and 62% with diagnosed hypertension.

The demographic and clinical measures by histology group are shown in Table 1. The mean age, BMI, and proportion of female subjects were similar across groups. The prevalence of diabetes, hypertension, hyperlipidemia, and National Cholesterol Education Program MS was also similar across histological groups. More white subjects trended to be present in the advanced liver histology groups (borderline and NASH) (P = 0.15).

Table 1.  Demographic variable by histology diagnostic groups
inline image

Comparison of metabolic variables by liver histological group

Based on ANOVA results, concentrations of HDL, TG, aspartate aminotransferase, and ALT were significantly different across groups (Table 2). Levels of HDL were significantly lower in NASH compared to no NAFLD (45 ± 11 mg/dl vs. 56 ± 14 mg/dl, P = 0.0027). Figure 1 demonstrates the dot plot of log-transformed TG concentrations by histological diagnosis groups. TG levels increased significantly (P = 0.0014) with progression of no NAFLD to simple steatosis, and from steatosis to definitive NASH. NASH and borderline groups had significantly greater TG than no NAFLD group (P = 0.0027) (Table 2). Subjects with definitive NASH by histology had twofold increased concentrations of ALT and AST (49 ± 42 U/l vs. 19 ± 10 U/l, P < 0.0001 and 45 ± 41 U/l vs. 21 ± 7 U/l, P < 0.0001, respectively). However, average ALT levels were in the normal reference range for no NAFLD, simple steatosis, and borderline categories. A dot plot of ALT levels by histological diagnosis groups is depicted in Figure 2.

Table 2.  Clinical and metabolic measures for liver histology diagnostic groups
inline image
image

Figure 1. Dot plot graph of log-transformed triglyceride (TG) levels by liver histology diagnostic groups. The dashed line represents log10(150), which is the metabolic syndrome diagnostic threshold for TG. NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis.

Download figure to PowerPoint

image

Figure 2. Dot plot graph of alanine transaminase (ALT) concentrations by liver histology diagnostic groups. The dashed line represents 45, which is the upper limit of normal reference range. Levels ≥45 U/l effectively identified nonalcoholic steatohepatitis (NASH) in our cohort by logistic analysis. NAFLD, nonalcoholic fatty liver disease.

Download figure to PowerPoint

Adipokine levels by liver histology groups

Serum RBP4, adiponectin, and leptin levels were not statistically significantly different across histological groups (Table 2). Levels of high-sensitivity TNF-α were similar between no NAFLD and NASH groups (2.2 ± 0.91 pg/ml vs. 1.96 ± 0.74 pg/ml, P = 0.17). Although mean adiponectin levels were higher in the no NAFLD group (7.690 ± 5.075 µg/ml) as compared to the NASH group (5.088 ± 3.013 µg/ml), the difference was a trend and did not meet statistical significance (P = 0.061). There were also no differences between histological groups for the ratio of TNF-α-to-adiponectin. No statistical differences were established between liver histology groups for subjects with T2DM or the MS.

Logistic regression models

To investigate the independent effects of these variables on liver disease, we conducted logistic regression analysis. We used two predictive models: NASH (vs. all other outcomes) and no NAFLD (vs. all other outcomes). After performing a stepwise logistic regression analysis for the first predictive model, we found that ALT and the binary TG variables (defined to be ≥150 mg/dl) predicted NASH. Individuals with TG >150 mg/dl had a 3.4-fold greater risk of developing NASH than those with TG <150 mg/dl (P = 0.0383), and a unit increase in ALT increased the risk of NASH by 6% (P = 0.0016). Using a binary cutoff for ALT as >45 U/l, best predicted NASH. After accounting for ALT and TG variables, HDL levels did not predict NASH. In the second predictive model, both HDL and ALT levels were significant in predicting no NAFLD. Controlling for ALT, a unit increase in HDL decreased the risk for liver disease (NASH, simple steatosis, or borderline diagnosis) by 5%. In other words, a unit decrease in HDL increased the chances of no disease by 5% (P = 0.0129). When controlling for HDL, a unit increase in ALT corresponded to an increase in risk for liver disease by 6% (P = 0.0113). TG levels, as defined earlier, did not predict the absence of NAFLD. Applying receiver operating characteristic analysis for prediction of NASH resulted in an area under the curve of 0.65 for ALT >45 U/l and 0.60 for TG >150 mg/dl.

Adipokine relationships with clinical variables

As expected, leptin levels strongly correlated with BMI (r = 0.41, P < 0.01) but not with other clinical variables. Adiponectin was positively correlated with concentrations of HDL (r = 0.34, P < 0.01) and inversely with TG levels (r = −0.26, P < 0.01). TNF-α associated with fasting glucose (r = 0.17, P = 0.04); serum RBP4 related with age (r = 0.26, P < 0.01), inversely with BMI (r = −0.27, P < 0.01), and linearly with TG levels (r = 0.18, P = 0.05); and TG levels were associated with RBP4 (r = 0.18, P = 0.05), fasting glucose (r = 0.41, P < 0.001) and inversely with HDL (r = −0.29, P < 0.01).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

The results of this study demonstrate that NAFLD is indeed frequent in an obesity surgery cohort with over two thirds demonstrating histological presence of NAFLD and 18% with definitive NASH by liver biopsy. There was a high prevalence of dyslipidemia, hypertension, and glucose intolerance but few variables associated with the National Cholesterol Education Program Adult Treatment Panel III defined MS identified advanced forms of NAFLD and NASH. Levels of TG strongly associated with histological severity of NAFLD, whereas increasing levels of HDL decreased likelihood of NASH. Moreover, ALT levels >45 U/l identified NASH in this cohort, with every one unit increase in ALT resulting in 6% increase in likelihood of NASH. Contrary to our expectation, there were no significant differences between the liver histology groups based on adipokine levels including RBP4, adiponectin, TNF-α as well as TNF-α/adiponectin ratio. However, there was a trend toward adiponectin levels to be decreased in the NASH vs. no NAFLD group (P = 0.061).

Our cohort, similar to those reported previously (1,5,6) presenting for obesity surgery was characterized by clinically severe obesity with BMI >35 kg/m2, primarily female gender, and high prevalence of obesity-related comorbid conditions including NAFLD. The prevalence of severe adiposity, T2DM, MS, and hypertension in our study were similar across histological groups and thus did not relate with advanced stages of NAFLD. However, the prevalence of T2DM may have been underestimated in this clinical cohort due to the lack of glucose tolerance testing. The prevalence of no NAFLD and advanced stages of NALFD are consistent with previous reports examining obesity surgery cohorts (1,5,6). In addition to prevalence of T2DM and MS components, previous reports have linked NAFLD to male gender, BMI, and abdominal adiposity and insulin resistance levels in obesity surgery cohorts (1).

Among the MS components, elevated TG concentrations in this cohort were strongly associated with severity of NAFLD and presence of NASH. Elevated TG and low HDL levels are characteristically found in individuals with insulin resistance, T2DM, and the MS (8,12,20), and these conditions have been closely related by others to progression of NAFLD to NASH (2). The impressive association between circulating TG concentrations and NASH underscores the importance of fatty acid delivery to the liver in the pathogenesis of NAFLD, in addition to the development of insulin resistance (9,21). Free fatty acids are the primary drivers of very low–density lipoprotein secretion in the liver (22). Thus, the same process that results in increased very low–density lipoprotein–TG secretion by the liver and hyperlipolysis of adipose tissue is likely responsible in large measures for NAFLD development. As dietary fat is an important source for delivery of free fatty acids to the liver, the demonstration of up to 20% disposal of dietary fat by liver in dogs suggests that the liver is a prominent destination for ingested dietary fat in the form of chylomicrons (23). The clinical end point of these pathways translates to elevated circulating TG concentrations which can lead to low levels of HDL. Thus the reciprocal relationship of TG and HDL found in insulin resistant states is clearly relevant to identification of NASH.

Consistent with previous reports (6,24), ALT levels strongly related with advanced liver disease in our cohort. Emerging data suggest that liver enzymes may predict incidence of diabetes (7), and thus serve as noninvasive markers of hepatic fat deposition, hepatic insulin sensitivity, and increased hepatic glucose production (4). However, serum ALT levels are nonspecific for steatosis as a cause for liver injury and have been reported previously to be discordant with histological findings indicating intermediate and advanced liver injury (1,25). Our study suggests that the presence of elevated ALT levels in the setting of elevated TG levels may assist the clinician to identify individuals likely to have NAFLD vs. no NAFLD.

This is the first examination of various circulating adipokines in relation to liver histology in patients with clinically severe obesity. Although levels of adiponectin, TNF-α, RBP4, and leptin levels related to glucose, lipids, and other MS components in our cohort, none of the adipokines measured related with presence of NAFLD or NASH. In vitro studies and experimental animal studies have proven the importance of inflammatory effects of TNF-α in inducing insulin resistance, accumulation of hepatic fat, and liver injury (2,10,13,16,26). Neutralization of TNF-α activity improved insulin resistance and fatty liver in these animals. However, our results suggest that TNF-α as well as other adipokines may be involved regionally in liver, but no global differences were detected in the circulation in a cohort of severely obese individuals with and without NAFLD. In contrast to TNF-α, interleukin 6 is synthesized by liver and was recently shown to predict NASH (27). Similarly, adiponectin has been demonstrated to have neutralizing effects on TNF, thus insulin sensitizing and anti-inflammatory (10,13,28,29), but these levels only trended to be different between no NAFLD and NASH groups (P = 0.061). The ratio of TNF-α-to-adiponectin levels did not differ between simple steatosis, borderline, and definitive NASH groups, suggesting that the ratio may not be of use in differentiating advance stages of NAFLD in obese individuals. Various drug treatments for T2DM, hypertension, and hyperlipidemia with thiazolidinediones, statins, ace inhibitors, biguanide, and insulin therapy have been shown to improve clinical severity of MS (2,30) and may alter adipokine levels measured in this cohort and thus blurred the distinction between histological groups. A randomized control trial using a thiazolidinedione, pioglitazone demonstrated reduction of circulating TG, free fatty acids, improvement in insulin sensitivity with diminution of liver steatosis, fibrosis, and inflammation (30).

RBP generated in the liver has been associated with development of insulin resistance in development of diabetes, obesity, and genetically predisposed offspring (31). However, our finding of lack of association with liver histology is consistent with previous literature (32). In a study of chronic liver disease, serum RBP4 was reduced in liver cirrhosis vs. healthy subjects with no liver disease, but serum RBP4 did not relate to insulin resistance associated with chronic liver disease (33). Elevated serum RBP4 has been linked to hepatic fat and viral hepatitis in nonobese, nondiabetic individuals (18,31). Our finding linked RBP4 to BMI and TG levels but not with liver histology, suggesting that these levels relate to insulin resistance indexes but not with severity of liver injury.

Several limitations must be acknowledged in this study. First, given the cross-sectional design of this study, we cannot conclude that clinical markers predict progression of liver disease, as this would imply examining these components prospectively in the same individuals. Second, given that this is a clinical cohort of obesity surgery subjects, the prevalence of medications known to impact NASH and metabolism could not be completely accounted for. The prevalence of T2DM and MS components was likely underestimated in the cohort because many individuals underwent medical weight loss strategies prior to presenting for surgery.

In summary, obesity is a dominant risk factor for developing NAFLD, and subjects with clinically severe obesity have a very high prevalence of NAFLD. Severely obese people with NAFLD are more likely to have advanced forms of NAFLD, identified largely by elevated TG and ALT levels. Contrary to our expectations, adipokines did not differentiate between benign and advanced liver histology stages and thus were not associated with NASH in this cohort.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

We are grateful to the skilled technical assistance of Rose Lounsbury for performing adipokine measurements and Steve Maximuk in the Preventive Cardiology Core Laboratory for processing serum samples. This work was supported by the National Institutes of Health, the National Center for Research Resources, Multidisciplinary Clinical Research Career Development Programs Grant no. 5K12RR023264 (C.M.S. and S.R.K.).

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES