Metabolic correlates of nonalcoholic fatty liver in women and men

Authors

  • Gloria Lena Vega,

    Corresponding author
    1. The Center for Human Nutrition of the University of Texas Southwestern Medical Center at Dallas, Texas
    2. Donald W. Reynolds Cardiovascular Research Center of the University of Texas Southwestern Medical Center at Dallas, Texas
    3. Departments of Clinical Nutrition and Internal Medicine of the University of Texas Southwestern Medical Center at Dallas, Texas
    4. The Veterans Affairs Medical Center, Dallas, TX
    • University of Texas Southwestern Medical Center, Center for Human Nutrition, 5323 Harry Hines Boulevard, Dallas, TX 75390-9052
    Search for more papers by this author
    • fax: 214-648-4837

  • Manisha Chandalia,

    1. The Center for Human Nutrition of the University of Texas Southwestern Medical Center at Dallas, Texas
    2. Donald W. Reynolds Cardiovascular Research Center of the University of Texas Southwestern Medical Center at Dallas, Texas
    3. Departments of Clinical Nutrition and Internal Medicine of the University of Texas Southwestern Medical Center at Dallas, Texas
    Search for more papers by this author
  • Lidia S. Szczepaniak,

    1. Donald W. Reynolds Cardiovascular Research Center of the University of Texas Southwestern Medical Center at Dallas, Texas
    2. Departments of Clinical Nutrition and Internal Medicine of the University of Texas Southwestern Medical Center at Dallas, Texas
    Search for more papers by this author
  • Scott M. Grundy

    Corresponding author
    1. The Center for Human Nutrition of the University of Texas Southwestern Medical Center at Dallas, Texas
    2. Donald W. Reynolds Cardiovascular Research Center of the University of Texas Southwestern Medical Center at Dallas, Texas
    3. Departments of Clinical Nutrition and Internal Medicine of the University of Texas Southwestern Medical Center at Dallas, Texas
    4. The Veterans Affairs Medical Center, Dallas, TX
    • University of Texas Southwestern Medical Center, Center for Human Nutrition, 5323 Harry Hines Boulevard, Dallas, TX 75390-9052
    Search for more papers by this author
    • fax: 214-648-4837


  • Potential conflict of interest: Nothing to report.

Abstract

Nonalcoholic hepatic steatosis associates with a clustering of metabolic risk factors and steatohepatitis. One risk factor for hepatic steatosis is obesity, but other factors likely play a role. We examined metabolic concomitants of hepatic steatosis in nonobese and obese men and women. Sixty-one obese women and 35 obese men were studied; both those with and without hepatic steatosis were compared against each other and against nonobese controls (17 women and 32 men) without hepatic steatosis. Obesity (defined as ≥25% body fat in men and ≥35% in women), was identified by x-ray absorptiometry, whereas hepatic steatosis (≥5.5% liver fat) was detected by magnetic resonance spectroscopy. The primary endpoint was a difference in insulin sensitivity. Obese groups with and without steatosis had similar body fat percentages. Compared with obese women without hepatic steatosis, those with steatosis were more insulin resistant; the same was true for men, although differences were less striking. Obese subjects with hepatic steatosis had higher ratios of truncal-to-lower body fat and other indicators of adipose tissue dysfunction compared with obese subjects without steatosis. Conclusion: These results support the concept that obesity predisposes to hepatic steatosis; but in addition, insulin resistance beyond that induced by obesity alone and a relatively high ratio of truncal-to-lower body fat usually combined with obesity to produce an elevated liver fat content. (HEPATOLOGY 2007.)

An increased hepatic triglyceride content (HTGC) associates commonly in the metabolic syndrome and can predispose to liver disease.1 Obesity is well recognized as 1 underlying cause.1, 2 Several mechanisms may play a role: high plasma nonesterified fatty acids (NEFA),3, 4 reduced adiponectin levels,5, 6 and increased fatty acid synthesis in the liver.7

A recent report1 on a representative population from Dallas Country [the Dallas Heart Study (DHS)] showed a high prevalence of hepatic steatosis. In DHS, HTGC correlated with body mass index, confirming a contribution from obesity. However, DHS findings also indicated that other yet-to-be-identified factors can raise HTGC. One factor may be insulin resistance,8, 9 but because obesity itself is accompanied by increased insulin resistance, the pathophysiological significance of the relationship between insulin resistance and hepatic steatosis is not clear. We therefore addressed whether systemic metabolic abnormalities in obese women and men with hepatic steatosis, compared with obese persons without steatosis at a similar percentage of total body fat, can be identified. The primary focus of the study was on insulin sensitivity differences; however, differences in adipose tissue metabolism were also examined.

Abbreviations

AUC, area under the curve; CRP, C-reactive protein; DHS, Dallas Heart Study; DXA, dual x-ray absorptiometry; HTGC, hepatic triglyceride content; MRS, magnetic resonance spectroscopy; NEFA, nonesterified fatty acids; OGIS, oral glucose insulin sensitivity index; OGTT, oral glucose tolerance test.

Patients and Methods

Experimental Design.

An elevation of HTGC was identified by magnetic resonance spectroscopy (MRS).1 The upper limit of normal (95th percentile) for HTGC by MRS in 345 DHS participants who were devoid of secondary causes of hepatic steatosis was ≤5.5% liver fat. This value was used to define normal and high HTGC in the current study.

The current subjects were selected based on HTGC and body fat as percentage of total body weight.1 HTGC was determined by MRS,1 and percent body fat and truncal-to-lower body fat ratios were measured as recently described by dual-x-ray absorptiometry (DXA).10 Figure 1 compares HTGC with percent body fat in women and men from the DHS database.1 At lower percentages of body fat, a high HTGC was uncommon; at higher percentages, a subgroup of persons developed a high HTGC. Based on these results, and MRS and DXA prescreening, subjects were recruited into 3 groups: nonobese with normal HTGC, obese with normal HTGC, and obese with high HTGC. Cutoff points to define obesity were based on the data of Jackson et al.11 from the Heritage Study; in this study, a body fat of 25% for men and 35% for women corresponds on average to a body mass index of approximately 27 kg/m2. These values for percent body fat were taken as the threshold to define obesity in the current study.

Figure 1.

Plot of liver fat percentage versus body fat measured by DXA in men (n = 1052) and women (n = 1189) of the Dallas Heart Study. Vertical lines divide nonobese and obese subjects at 25% total body fat for men and 35% for women. Horizontal lines divide normal and high hepatic triglyceride content at 5.5%.

Subjects.

Subjects were excluded if they had a history of alcohol ingestion (defined by >28 g/day in men and >14 g/day in women) within the past 6 months of enrollment, a fasting plasma glucose ≥126 mg/dl, a thyroid-stimulating hormone outside the normal range, a creatinine ≥1.4 mg/dl, and a physical illness that in the judgment of the investigators would disqualify the subject from the study. Patients were selected to have AST and ALT below the upper limit of the references ranges, 10-42 IU/l and 10-40 IU/l, respectively. Mean ASTs and ALTs for the group were 22 ± 8 IU/l (SD) and 21 ± 10 IU/l, respectively. Chronic hepatitis was excluded by history and normal liver function tests. The exclusion of subjects with raised AST and ALT leads to partial exclusion of subjects with nonalcoholic steatohepatitis, but mostly in the male group, because of the gender-specificity of the cutoff value. However, some subjects with normal AST/ALTs might still have nonalcoholic steatohepatitis. Mean creatinine levels ranged from 0.6 to 1.3 mg/dl.

Other exclusion criteria included hypolipidemic drugs, steroids, unstable hypertension, heavy smoking, and dietary supplements of vitamin E, nicotinic acid, or stanol esters. Subjects with a history of coronary heart disease, pacemakers, significant chronic diseases, or who were pregnant were excluded. Our Institutional Review Board approved the study.

Invitation for participation was made on baseline DXA and MRS data from DHS. Ninety-six percent of the recruited subjects were derived from the original DHS cohort. The remaining subjects came through new screening. Because we recruited a convenience sample and it had a nonhomogeneous ethnic distribution, the results cannot necessarily be generalized to the whole population. All selection for the current study was based on either repeat DXA and MRS or new measurements. Seventy-eight women and 67 men (ages 18-69 years) were recruited. Their distributions according to body weight category and liver fat content are presented in the Results section. Thirty-two women and 22 men were self-reported black; the remainder were non-Hispanic white. Nineteen subjects had stable hypertension; 51 women and 46 men had either impaired fasting glucose or impaired glucose tolerance but none had type 2 diabetes.

Metabolic Studies.

The subjects underwent an extended glucose tolerance test in the General Clinical Research Center. After a 12-hour fast, at 8:00 AM, 2 blood samples were obtained over 30 minutes. Then each subject was given 75 g glucose orally. Arterialized blood samples were drawn over 5 hours. Glucose and NEFA were measured enzymatically, c-peptide and insulin by immunoassay, and baseline leptin and adiponectin by radioimmunoassay. Blood chemistries and plasma lipids were also measured.10 Two models of insulin sensitivity were used: oral glucose insulin sensitivity index (OGIS) of Mari et al12 and insulin sensitivity index of Breda et al.13 The results of both methods give a relatively high correlation with glucose disposal estimated by the glucose-clamp technique.

Statistical Methods.

The primary endpoint for the study was insulin sensitivity to glucose regulation by the OGIS method in obese individuals with and without hepatic steatosis. For the power calculation, we assumed a 20% difference between groups (obese without hepatic steatosis and with hepatic steatosis) and 20% standard deviation, an alpha of 0.05, a power >0.95. Therefore, we estimated approximately 20 subjects per cell (liver content and sex). Descriptive statistics were computed for each subgroup of sex and low/high liver fat groups. A 2-way ANOVA model was used to assess possible differences between normal HTGC and high HTGC groups. Multiple comparisons between the normal and high HTGC subjects for each sex and body-fat category also were assessed. Statistical analyses were carried out using the statistical package SPSS (12.0.1) from SPSS Inc. (Chicago, IL).

Results

Insulin Sensitivity.

Indicators of insulin sensitivity are shown for women and men in Table 1. In women, fasting glucose, insulin, c-peptide, and the AUC for insulin during the oral glucose tolerance test (OGTT) were progressively and significantly higher in nonobese, obese with normal HTGC, and obese with high HTGC. Conversely, during OGTT, OGIS and insulin sensitivity index were progressively lower. In men, fasting insulin and c-peptide were statistically higher in obese with high HTGC compared with obese with normal HTGC. The same pattern was observed but conversely for OGIS, with a trend for lowest values for insulin sensitivity index in the obese with steatosis.

Table 1. Insulin Sensitivity Parameters
ParameterWomen
Nonobese [n = 17]Obese (Normal HTGC) [n = 37]Obese (High HTGC) [n = 24]
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
  • *

    Significantly higher than nonobese group (P <0.05).

  • Significantly higher than obese group with normal HTGC (P <0.05).

  • Data were loge transformed for analysis of variance.

  • Abbreviations: AUC, area under the curve; OGIS, oral glucose insulin sensitivity.

Fasting blood glucose (mmol/l)5.0 ± 0.44.95.2 ± 0.45.25.7 ± 0.7*5.6
Fasting insulin (pmoles/l) 61 ± 445079 ± 34*73133 ± 78*118
Fasting c-peptide (pmoles/l) 522 ± 163509652 ± 243586992 ± 371*941
AUC insulin63515 ± 875151274116059 ± 28136*76680134405 ± 11849*128768
AUC c-peptide354302 ± 22895367959431105 ± 19705434186551031 ± 37268*527902
OGIS 180′ (ml/min/m2)454 ± 85454371 ± 65*364312 ± 64*314
Insulin sensitivity index (dl/min/kg/pmole/l)12.17 ± 8.3310.817.07 ± 3.9*7.344.06 ± 3.55*2.58
 Men
Nonobese [n = 32]Obese (Normal HTGC) [n = 18]Obese (High HTGC) [n = 17]
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
Fasting blood glucose (mmol/l)5.3 ± 0.45.35.6 ± 0.45.45.5 ± 0.5*5.4
Fasting insulin (pmoles/l)59 ± 275989 ± 44*74105 ± 34*96
Fasting c-peptide (pmoles/l)621 ± 270544804 ± 4127421078 ± 515*939
AUC insulin50073 ± 46814396684957 ± 14028*64712105606 ± 9756*97664
AUC c-peptide338203 ± 18402331460436237 ± 21924*430756523275 ± 49139*471076
OGIS 180′ (ml/min/m2)440 ± 71426387 ± 77*411342 ± 63*335
Insulin sensitivity index (dl/min/kg/pmole/l)16.3 ± 13.9812.447.78 ± 4.23*5.836.84 ± 6.17*3.95

Body Fat Relationships.

For both men and women, HTGC was distributed according to recruitment criteria (Table 2). For the 2 obese groups of both sexes, body fat as a percentage of total body mass (% body fat) also was higher in accord with recruitment criteria; but among obese groups without or with increased HTGC, no differences in percent body fat were present. In women, waist circumferences and percent truncal fat were higher in both obese groups compared with nonobese, but in obese women with high HTGC, waist circumference and percent truncal fat were significantly higher than in the obese with normal HTGC. Percent lower body fat was greater in both groups of obese women, but it was significantly lower in obese women with high HTGC. Consequently, the ratio of truncal-to-lower body fat was significantly higher in obese women with high HTGC than in those with normal HTGC. Among obese men, those with high HTGC had a nonsignificant trend toward higher waist circumferences and percent truncal fat and toward a lower percent lower body fat. However, the ratio of truncal-to-lower body fat was significantly higher in obese men with high HTGC compared with those with normal HTGC.

Table 2. Body Composition in Nonobese and Obese Women and Men With Low and High Hepatic Triglyceride Content (HTGC)
ParameterWomen
NonobeseObese (Normal HTGC)Obese (High HTGC)
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
  • *

    Significantly higher than nonobese group (P <0.05).

  • Significantly higher than obese group with normal HTGC (P <0.05).

  • ‡Data were loge transformed for analysis of variance.

Hepatic triglyceride content (%)c1.9 ± 1.01.82.4 ± 1.41.915.0 ± 6.0*14.4
Percent body fat (% of total mass)27.1 ± 4.428.639.6 ± 4.1*39.542.3 ± 5.0*41.4
Waist circumference (cm)79.4 ± 8.380.097.1 ± 14.4*92.7114.8 ± 17.5*114.3
Truncal fat (% of total mass)10.0 ± 4.111.017.0 ± 3.2*16.420.7 ± 3.6*20.5
Lower body fat (% of total mass)12.4 ± 1.612.816.7 ± 2.1*16.315.3 ± 2.4*14.7
Ratio of truncal fat-to-lower body fat0.81 ± 0.340.791.03 ± 0.23*1.041.37 ± 0.26*1.41
 Men
NonobeseObese (Normal HTGC)Obese (High HTGC)
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
Hepatic triglyceride content (%)c1.9 ± 1.01.92.7 ± 1.02.69.8 ± 3.510.7
Percent body fat (% of total mass)18.9 ± 4.620.328.4 ± 2.6*28.629.6 ± 4.9*29.4
Waist circumference (cm)90.2 ± 8.990.8105.6 ± 7.2*104.8114.3 ± 18.8*106.7
Truncal fat (% of total mass)9.0 ± 3.49.814.0 ± 2.0*13.916.3 ± 3.5*16.3
Lower body fat (% of total mass)6.3 ± 1.86.89.7 ± 1.1*9.79.1 ± 1.8*9.0
Ratio of truncal fat-to-lower body fat1.48 ± 0.51.421.46 ± 0.31.401.84 ± 0.43*1.89

Fasting levels of NEFA, leptin, adiponectin, and C-reactive protein (CRP) levels are given for women and men in Table 3; also shown are area under the curve (AUC) for NEFA during the OGTT. In women, mean fasting NEFA were progressively and significantly higher in nonobese, obese with normal HTGC, and obese with high HTGC. The AUC for NEFA during OGTT was significantly higher in obese women with high HTGC compared with the other 2 groups. Leptin and CRP levels were higher in obese women, but there were no differences between the 2 obese groups. In contrast, adiponectin levels were significantly lower in obese women with high HTGC than in those with normal HTGC. In men, similar but less significant trends noted in women were present. Although NEFA were not higher in obese men with high HTGC, the AUC for NEFA during OGTT was significantly higher in these men than in the other 2 groups. Leptin and CRP levels were similar in both obese groups and higher than in nonobese. But only in obese men with high HTGC were mean adiponectin levels lower than in the nonobese.

Table 3. Adipose Tissue Products
ParameterWomen
NonobeseObese (Normal HTGC)Obese (High HTGC)
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
  • *

    Significantly higher than nonobese group (P <0.05).

  • Significantly higher than obese group with normal HTGC (P <0.05).

  • Data were loge transformed for analysis of variance.

  • Abbreviations: NEFA, nonesterified fatty acids; AUC, area under the curve.

NEFA levels (μmol/l)444 ± 198421512 ± 155*520538 ± 169*549
NEFA AUC × 10323.8 ± 8.023.429.3 ± 8.7*27.635.9 ± 11.4*35.0
Leptin (ng/ml)68.4 ± 40.163.3209.0 ± 118.5*190.0277.6 ± 111.1*246.0
Adiponectin (μg/ml)15.6 ± 6.814.014.0 ± 7.012.710.3 ± 6.8*8.5
C- reactive protein (mg/l)1.7 ± 2.60.95.2 ± 9.1*2.84.9 ± 3.1*4.7
 Men
NonobeseObese (Normal HTGC)Obese (High HTGC)
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
NEFA levels (μmol/l)417 ± 235393348 ± 124*336465 ± 152*464
NEFA AUC × 10323.7 ± 8.324.424.8 ± 6.522.633.9 ± 12.1*33.8
Leptin (ng/ml)25.0 ± 12.626.468.5 ± 29.4*53.492.9 ± 57.5*66.6
Adiponectin (μg/ml)13.6 ± 7.011.89.6 ± 4.8*8.09.0 ± 3.8*7.7
C- reactive protein (mg/l)‡2.2 ± 4.20.88.1 ± 23.0*1.88.3 ± 22.1*2.0

By multiple regression analysis for the whole groups of men and women, percent total body fat alone explained 13.4% of the variance in liver fat content in women and 33.5% in men. In men, the combination percent body fat + body fat ratios (truncal to lower body fat ratios) explained 35.4% of variance, and by adding OGIS, 36.8%. In all women, adding body fat ratios to percent body fat accounted for 28.7% of variance, and with addition of OGIS, 34.9%. Within the obese category, percent total body fat alone explained only 3.7% of the variance in liver fat content in women and 14.9% in men. In obese men, the combination percent body fat + body fat ratios explained 21% of variance, and by adding OGIS, 28.6%. In obese women, adding body fat ratios to percent body fat accounted for 28.2% of variance, and by addition of OGIS, 38.8%.

Comparison of key parameters between black and white women and black and white men showed essentially no ethnic differences between any of the 3 groups (Table 4).

Table 4. Ethnic Comparison for Key Parameters* (Mean ± SD)
ParameterEthnicityWomenMen
Non-ObeseObese (Normal HTGC)Obese (High HTGC)Non-ObeseObese (Normal HTGC)Obese (High HTGC)
  • NOTE. Non-Obese: black women, n = 8; white women, n = 9; black men, n = 12, white men, n = 20

  • Obese (Normal HTGC): black women, n = 13; white women, n = 24; black men, n = 5, white men, n = 13

  • Obese (High HTGC): black women, n = 11; white women, n = 13; black men, n = 5, white men, n = 12

  • *

    If no symbol by comparison, differences were not statistically significant by unpaired t test. †Significantly different by unpaired t test; P = 0.02.

Liver fat content (%)Black1.7 ± 0.82.6 ± 0.913.7 ± 5.72.1 ± 0.093.2 ± 1.18.3 ± 5.0
 White2.0 ± 1.12.1 ± 1.316.1 ± 6.21.8 ± 1.02.6 ± 1.010.4 ± 2.8
OGIS (ml/min/m2)Black412 ± 89346 ± 75296 ± 44436 ± 76294 ± 58314 ± 35
 White492 ± 66384 ± 56325 ± 77443 ± 69419 ± 49354 ± 71
Si (ml/min/kg/pmol/l) × 105Black10.4 ± 7.85.8 ± 3.62.3 ± 0.9†9.1 ± 7.55.8 ± 4.76.1 ± 8.3
 White13.8 ± 8.97.8 ± 3.95.5 ± 4.320.6 ± 15.38.1 ± 4.17.5 ± 5.6
c-peptide (pmol/l)Black548 ± 163645 ± 2581000 ± 167708 ± 246815 ± 1141201 ± 717
 White499 ± 169656 ± 241986 ± 490571 ± 277800 ± 4861027 ± 435
Adiponectin (μm/ml)Black14.3 ± 5.111.3 ± 5.17.4 ± 2.411.5 ± 5.46.5 ± 2.38.2 ± 2.1
 White16.7 ± 8.215.5 ± 7.512.8 ± 8.414.8 ± 7.610.8 ± 4.99.4 ± 4.4
Truncal-to-lower body fat ratioBlack10.6 ± 3.417.2 ± 3.220.0 ± 4.09.2 ± 4.612.8 ± 2.017.7 ± 6.4
 White9.4 ± 4.816.9 ± 3.321.3 ± 3.28.9 ± 2.714.7 ± 1.815.7 ± 2.1

Discussion

Almost one-third of the whole adult population in the DHS had hepatic steatosis.1 DHS data further suggested that obesity is virtually a requisite for development of hepatic steatosis (Fig. 1); yet, metabolic differences presumably exist among obese subjects that lead to steatosis in some but not others. In this study we addressed whether systemic metabolic differences can be detected between obese persons with and without hepatic steatosis. Among white and black men and women, within a given body-fat category, no ethnic differences were observed for metabolic correlates. The metabolic focus of our study was on insulin sensitivity. The major finding was that those obese persons with hepatic steatosis are less insulin sensitive than obese persons with normal liver fat content even when both groups had a similar percent total body fat. This finding suggests a dual defect underlying hepatic steatosis, namely, obesity plus a second metabolic defect. The latter appears to be linked in some way to a greater insulin resistance than occurs with obesity alone.

Insulin Resistance and Hepatic Steatosis.

That persons with hepatic steatosis commonly are insulin resistant is known.8, 9, 13 However, a given degree of obesity individuals with hepatic steatosis are more insulin resistant than are those without. Our study shows that obese subjects with steatosis in fact are more insulin resistant than those without. It further suggests that this greater insulin resistance originates in adipose tissue. Subjects with hepatic steatosis had either higher NEFA levels or AUCs of NEFA during OGTT; higher NEFA, which should raise liver fat content, impairs insulin signaling in muscle and liver.14–16 Moreover, obese women with steatosis had lower adiponectin levels than those without; these lower levels could reduce fatty acid oxidation and raise liver-fat content.6 Furthermore, hyperinsulinemia, secondary to insulin resistance and high NEFA levels, could enhance fatty acid synthesis and inhibit fatty acid oxidation in the liver15, 16; both should increase HTGC and exacerbate insulin resistance.17, 18 Finally, inflammatory cytokines released by adipose tissue may further accentuate insulin resistance.19 Whether obese subjects with hepatic steatosis had higher levels of circulating cytokines was not examined in this study. No differences were observed between CRP levels between obese individuals with and without steatosis, but CRP levels may not be sensitive enough to identify differences in cytokine levels. All told, obese subjects with steatosis had several abnormalities suggestive of a dysfunctional adipose tissue that can explain higher insulin resistance and increased liver fat content. Increased insulin resistance itself likely signifies the presence of multiple metabolic abnormalities that can affect hepatic fat content.

Obesity Plus Adipose Tissue Disorders as Cause of Hepatic Steatosis.

If dysfunctional adipose tissue combines with obesity to produce hepatic steatosis, what might be the disorder? Two readily identified disorders are adipose tissue deficiencies (lipodystrophies) and body-fat misdistribution characterized by upper body fat. Lipodystrophies commonly manifest fatty liver.20 Upper body obesity may in fact represent a form of adipose-tissue deficiency in which there is an inadequate reservoir of lower body adipose tissue.10 When lower body adipose tissue is deficient, overloading with upper body fat may lead to ectopic fat accumulation. Upper body obesity, particularly of the visceral type, appears to be more commonly associated with fatty liver than is lower body obesity.21 A third type of disorder is a reduced adipose-tissue sensitivity to insulin.22 Although obesity per se induces insulin resistance of adipose tissue, other defects may enhance resistance beyond that observed with obesity alone. Insulin-resistant South Asians appear to have a dysfunctional adipose tissue in that they have the metabolic characteristics of obesity even in the absence of obesity, i.e., elevated NEFA, elevated CRP, and reduced adiponectin levels.23, 24 Current findings point to several defects in adipose tissue distribution or function in obese patients with hepatic steatosis compared with obese without steatosis. In obese women with steatosis, waist circumferences and percent truncal fat were higher than in obese women without steatosis, whereas percent lower body fat was lower. Consequently, the ratio of truncal-to-lower body fat was highest in obese women with steatosis. Furthermore, in this group, fasting NEFA levels were highest, and the AUC of NEFA during OGTT remained highest; conversely, adiponectin levels were significantly lower. In men with hepatic steatosis, the ratio of truncal-to-lower body fat was significantly higher than that of the other groups, and the AUC of NEFA during OGTT likewise was higher.

Protective Effect of Lower Body Fat Against Hepatic Steatosis.

Previous studies show that women with predominant upper body (truncal) obesity are more prone to insulin resistance and to type 2 diabetes than those with lower body obesity25, 26; we observed that they are more prone to hepatic steatosis and readily demonstrable metabolic differences. Thus a pattern of lower body fat seems protective. Men typically have a predominance of truncal fat even when not obese (Table 2); in fact, even nonobese men of the current study had the same ratios of truncal-to-lower body fat as obese women with hepatic steatosis. For this reason, the metabolic difference between obese men with and without steatosis likely was less distinct than that observed in women. Presumably men in general are prone to hepatic steatosis when they become obese because they generally are restricted to fat accumulation in the upper body.1

In summary, only a portion of obese persons develop hepatic steatosis. Therefore, a second abnormality must combine with obesity to promote triglyceride accumulation inliver. We attempted to uncover evidence for a systemic abnormality in persons with hepatic steatosis. Our primary evidence for this systemic abnormality was greater insulin resistance. Both obese women and men with steatosis were more insulin resistant than obese counterparts without steatosis. Both sexes who developed hepatic steatosis with obesity had evidence of a greater maldistribution of excess fat, which was best characterized by higher truncal-to-lower body fat ratios. However, the metabolic consequences of maldistribution were more readily detected in women than in men. Because men commonly have predominant upper body fat when obesity develops, the metabolic abnormalities responsible for hepatic steatosis are more subtle and more difficult to detect. Nonetheless, as shown in this study, their presence is revealed by a greater insulin resistance in those with hepatic steatosis.

Acknowledgements

The authors express their appreciation for the excellent clinical assistance of Nives Champion, PAC, Laura Caldwell, PAC, Marjorie Whelan, RN, and the nursing staff of the General Clinical Research Center. The technical assistance of Anh Nguyen, Biman Pramanik, Lindsay Blair, and Timothy Yates is gratefully acknowledged. Beverley Huet-Adams, Biostatistician of the General Clinical Research Center, was a consultant in the statistical analysis and study design.

Ancillary