Metabolic Significance of Nonalcoholic Fatty Liver Disease in Nonobese, Nondiabetic Adults

Background  Obesity and type 2 diabetes are well-known risk factors for the development of nonalcoholic fatty liver disease (NAFLD). However, NAFLD is not rare in nonobese, nondiabetic adults. The aim of this study was to evaluate the metabolic significance of NAFLD in nonobese, nondiabetic adults.

Methods  This study examined 768 nonobese (body mass index [BMI] [calculated as weight in kilograms divided by the square of height in meters], ≥18.5 and <30) (460 normal-weight and 308 overweight subjects), nondiabetic individuals older than 30 years who participated in a medical checkup. All the subjects had negative serologic findings for hepatitis B and C viruses and had an alcohol intake less than 140 g/wk. A standard interview, anthropometrics, a biochemical study, and abdominal ultrasonography were conducted.

Results  The prevalence of NAFLD in the enrolled subjects was 23.4%. In the normal-weight (BMI, ≥18.5 and <25) and overweight (BMI, ≥25 and <30) groups, NAFLD was a significant predictor of insulin resistance and other metabolic disorders, including hypertriglyceridemia and hyperuricemia. The odds ratio of the metabolic disorders in subjects with NAFLD compared with those without NAFLD in the normal-weight groupwashigher than that in the overweight group. Multiplelogisticregression analysis showed that sex, waist circumference, triglyceride level, and insulin resistance wereindependentlyassociated with NAFLD in the normal-weight group.

Conclusions  Nonalcoholic fatty liver disease is closely associated with metabolic disorders, even in nonobese, nondiabetic subjects. Nonalcoholic fatty liver disease can be considered an early predictor of metabolic disorders, particularly in the normal-weight population.

Nonalcoholic fattyliver disease (NAFLD)describes a clinicopathologic condition that is characterized by significant lipid deposition in the hepatocytes of the liver parenchyma in patients with no history of excessive alcohol consumption. The spectrum of this disease is broad, ranging from a simple steatosis to nonalcoholic steatohepatitis, fibrosis, and cirrhosis.^1-3

Nonalcoholic fatty liver disease has attracted a great deal of attention since the report by Leevy⁴ described 270 patients with NAFLD in 1962. However, the clinical implications of this condition are still under investigation. Nonalcoholic fatty liver disease is the most common liver disease in Western countries, and it is becoming increasingly prevalent in Asian-Pacific regions because of the increasing westernization of the lifestyle, such as a high-fat and high-calorie diet, less physical activity, and increasing incidence of central obesity and type 2 diabetes.⁵

It has been reported that NAFLD is related to obesity,^6,7 diabetes mellitus,^6,8,9 and dyslipidemia.^8-10 However, NAFLD has been found in individuals without such risk factors.⁹ Recent studies have suggested that hyperinsulinemia and insulin resistance may play a role in the pathogenesis of NAFLD.¹¹ It has now been proposed that NAFLD is associated with a metabolic syndrome.^12,13

Obesity and diabetes are well-known risk factors for the development of a fatty liver,⁸ and few studies have examined NAFLD in nonobese, nondiabetic populations. Although Asians are less obese than Western people, the prevalence of NAFLD and metabolic syndrome is not lower than that in Western people.¹⁴

Therefore, the aims of this study were to evaluate the metabolic significance of NAFLD in nonobese, nondiabetic adults and to assess the independent factors associated with NAFLD.

From April 1 to June 30, 2001, 932 individuals 30 years and older who received a medical checkup in the Korea Association of Health Promotion Center in Seoul, Korea, underwent screening.

The exclusion criteria were as follows: (1) alcoholic intake of 140 g/wk or more¹⁵; (2) National Institutes of Health–defined body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters) obesity (BMI, ≥30) or underweight (BMI, <18.5)¹⁶ according to the National Institutes of Health–defined cutoff point; (3) a history of diabetes mellitus or fasting hyperglycemia according to the American Diabetes Association criteria (>126 mg/dL [7.0 mmol/L])¹⁷; (4) a positive serologic finding for hepatitis B or C virus; (5) the presence of autoantibodies indicative of autoimmune hepatitis; (6) a history of another known liver disease; (7) a malignancy; (8) previous gastrointestinal tract surgery; and (9) ingestion of drugs known to produce hepatic steatosis, including corticosteroids, high-dose estrogens, methotrexate, tetracycline hydrochloride, amiodarone, or tamoxifen citrate in the previous 6 months.

Finally, 768 subjects were enrolled in this study. The Ethics Committee of Yonsei University College of Medicine, Seoul, approved this study, and informed consent was obtained from each subject.

All participants were interviewed to obtain their history of diabetes mellitus, hypertension, myocardial infarction, cerebrovascular accident, etc, as well as alcohol consumption and smoking. Their height, weight and waist and hip circumferences were measured to the nearest half-centimeter or half-kilogram. The waist circumference was measured at the midpoint between the lower border of the rib cage and the iliac crest, whereas the hip circumference was obtained at the widest point between the hip and buttock.

The blood pressure was measured using a standard mercury sphygmomanometer after the subject had been seated for at least 10 minutes. The percentage of whole body fat was measured by means of a bioelectrical impedance analysis method (Inbody 2.0; Biospace Co, Ltd, Seoul, Korea).

The laboratory evaluation included measurement of the aspartate aminotransferase, alanine aminotransferase, fasting blood glucose, fasting insulin, total cholesterol, high-density lipoprotein cholesterol (HDL-C), triglyceride, and uric acid levels. The low-density lipoprotein cholesterol level was calculated using the equation by Friedewald et al.¹⁸ The level of fasting blood glucose was determined by calculating the mean of the repeated checked value. The index of insulin resistance was calculated according to the homeostasis model assessment method (HOMA_IR),¹⁹ as follows:

Fasting Insulin Level×Fasting Glucose Level/22.5.

Similarly, the index of insulin secretion (HOMA_{β-cell function}) was calculated as

(20×Fasting Insulin Level)/(Fasting Glucose Level−3.5).

For both formulas, insulin was measured as microunits per milliliter; glucose, as millimoles per liter.

One experienced radiologist performed abdominal ultrasonography, and the presence of fatty liver was defined as an increased echogenicity of the hepatic parenchyma with an attenuation of the portal vein or diaphragm echogenicity.²⁰

The nonobese subjects were classified into the normal-weight (BMI, ≥18.5 and <25) and the overweight (BMI, ≥25 and <30) groups,¹⁶ and the subjects with and without NAFLD were compared within each BMI category.

Men and women with waist circumference values of at least 90 and 80 cm, respectively, were considered to have central obesity, according to the World Health Organization perspectiveon the western Pacific region for Asians.²¹ Hypertension was defined as a systolic blood pressure of 140 mm Hg or greater, a diastolic blood pressure of 90 mm Hg or greater, or the use ofantihypertensive medication, according to the guidelines of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.²² Impaired fasting glucose level was defined as a fasting plasma glucose level of 110 mg/dL or greater (≥6.1 mmol/L) but less than 126 mg/dL (<7.0 mmol/L). The presence of insulin resistance was defined as the highest quartile of the HOMA_IR.²³ Hypertriglyceridemia and low HDL-C level were defined according to the Adult Treatment Panel III of the National Cholesterol Education Program guidelines (triglyceride level, ≥150 mg/dL [≥1.7 mmol/L]; HDL-C level, <40 mg/dL [<1.0 mmol/L] in men or <50 mg/dL [<1.3 mmol/L] in women).²⁴ Hyperuricemia was defined as a uric acid level of 6.7 mg/dL or greater (≥400 μmol/L).¹²

The difference between groups with and without NAFLD was compared using an unpaired t test. The χ² test or Fisher exact test was used to compare the prevalence data. The odds ratio was calculated for 2×2 cross tables and is expressed with the 95% confidence interval. Logistic regression analysis was used to examine the independent factor on NAFLD. A 1-way analysis of variance (ANOVA) was used to assess the prevalence of a fatty liver according to the BMI, body fat, waist circumference, triglyceride level, and HOMA_IR quartile groups. When found to be significant, a post hoc (Scheffé method) multiple comparison was used to establish the differences between the groups. The statistical analyses were performed using the SPSS version 11.0 software package (SPSS Inc, Chicago, Ill). P values of less than.05 were considered significant.

The prevalence of a fatty liver was 23.4% in all the subjects and was higher in the overweight group than in the normal-weight group (34.4% vs 16.1%; P<.001).

In the clinical and laboratory data of all the enrolled subjects, there were significant differences between the subjects with and without NAFLD in the variables, including the BMI, body fat, waist circumference, lipid profile, HOMA_IR, and HOMA_{β-cell function} (Table 1).

The characteristics of the subjects according to the BMI are shown in Table 2. In the normal-weight group, there were significant differences between the subjects with and without NAFLD in terms of sex; BMI; waist circumference; waist-hip ratio; levels of total cholesterol, triglycerides, HDL-C, uric acid, fasting blood glucose, insulin, aspartate aminotransferase, and alanine aminotransferase; HOMA_IR; and HOMA_{β-cell function}. In the overweight group, a comparison between subjects with and without NAFLD showed a similar tendency to that in the normal-weight group (Table 2).

There were no significant differences between the normal-weight group with NAFLD and the overweight group without NAFLD in terms of the metabolic variables such as waist-hip ratio, blood pressure, lipid profiles, levels of fasting blood glucose and insulin, HOMA_IR, and HOMA_{β-cell function} (Table 2).

The prevalence and odds ratios of the metabolic disorders according to the presence of NAFLD are shown in Table 3 and Table 4, respectively. In the normal-weight group, the prevalence of hypertriglyceridemia, hyperuricemia, central obesity, and insulin resistance was higher in the subjects with a fatty liver than in those without one. However, in the overweight group, there was no difference in the prevalence of central obesity between the subjects with and without a fatty liver (Table 3).

In both the normal-weight and overweight groups, NAFLD was a significant predictor of insulin resistance and other metabolic abnormalities, including triglyceridemia and hyperuricemia. The odds ratios in the normal-weight group were higher than those in the overweight group (Table 4).

According to multiple logistic regression analysis,ageand BMI were independently associated with NAFLD in the overweight group, but not in the normal-weight group. In the normal-weight group, sex, waist circumference, triglyceride level, and logarithmHOMA_IRwere independently associated with NAFLD (Table 5).

The Figure shows the prevalence of NAFLD according to the quartiles of the factors associated with NAFLD. In the prevalence of NAFLD according to BMI quartiles, there were significant differences in both the normal-weight and overweight groups according to the ANOVA results, although not every quartile group showed a significant difference (Figure, A). However, the prevalence of NAFLD according to the body fat quartiles did not show any significant difference among the quartile groups according to ANOVA (data not shown). In terms of the waist circumference and triglyceride quartiles, the prevalence of NAFLD was significantly higher in the upper quartile groups in both the normal-weight and overweight groups (Figure, B and C). In the analysis according to theHOMA_IRquartiles, the prevalence of NAFLD was higher in the upper quartile groups (Figure, D).

Recently, an increasing number of studies have examined the significance of NAFLD, as NAFLD is increasingly prevalent and understood as a feature of the metabolic syndrome.²⁵ In the general population studies, screening with ultrasonography^26-29 or computed tomography³⁰ has indicated a prevalence ranging from 13% to 23%. In the present study, the prevalence of NAFLD in nondiabetic, nonobese adults was 23.4% (16.1% in the normal-weight group and 34.4% in the overweight group). Although the subjects in this study were nondiabetic, nonobese adults, the prevalence of NAFLD was higher than that in other reports in which the subjects were obtained from the general population. This suggests that lifestyle or genetic factors besides obesity, diabetes mellitus, and dyslipidemia may be a part of the risk factors for NAFLD. A recent study suggests that the fatty liver group had much greater carbohydrate intake than the control group and that the restriction of carbohydrates might contribute to the recovery of a fatty liver.³¹ Koreans have a higher carbohydrate intake than white subjects,^32,33 and the high prevalence of NAFLD can be partially ascribed to the increased intake of a higher percentage of carbohydrates in the regular diet.

In the present study, the diagnosis of NAFLD was based on the exclusion of the known etiologic factors responsible for liver disease and ultrasound examination results, but the diagnosis was not confirmed by liver biopsy results. In a prospective study comparing ultrasound scanning with histological examination results, Saverymuttu et al³⁴ showed that ultrasound examinations can accurately identify a steatosis with a sensitivity of 94% and a specificity of 84%. Although ultrasonography has some limitations in distinguishing a fatty liver from other liver diseases, the present study used ultrasonography to examine subjects in sufficient numbers, using a noninvasive method.

Diabetes and obesity are well-known risk factors for the development of NAFLD, but the causality of them to NAFLD is unclear. They may play an important role as confounding factors. Therefore, the significance of NAFLD, excluding diabetes and obesity, was investigated in this study.

Recent studies^11,12,35 have suggested that insulin resistance might be a primary phenomenon adding to obesity- and diabetes-associated insulin resistance in NAFLD. However, previous studies lacked an appreciation of NAFLD as a risk factor for metabolic disorders and did not evaluate NAFLD by comparing a normal-weight with an overweight group. This study assessed the NAFLD from the viewpoint of metabolic disorders, dividing the subjects on the basis of BMI. The results showed that NAFLD is a significant predictor of insulin resistance, and insulin resistance was found to be an independent factor associated with NAFLD on multiple logistic regression analysis in both the normal-weight and overweight groups. Not only was insulin resistance more prevalent in the subjects with NAFLD in both groups, also other metabolic disorders such as hypertriglyceridemia and hyperuricemia also were more prevalent.

The relationship between obesity and NAFLD is well known. However, it has been proposed that the waist-hip ratio or the waist circumference, which reflects central obesity, is more related to NAFLD than BMI is.^12,36 In contrast to the overweight group, the BMI was not an independent risk factor for NAFLD in the normal-weight group in the present study. However, patients with NAFLD showed an increased central obesity, even in the presence of a normal weight, and waist circumference was an independent risk factor for NAFLD, particularly in the normal-weight group. In addition, we found significant differences in the normal-weight and overweight groups in the prevalence of NAFLD according to the waist circumference quartiles by ANOVA. A previous study reported that a central body fat distribution determined by the waist-hip and waist-height ratios and the skinfold thickness predicted a fatty liver only in women.³⁷ However, in this study, when the data were analyzed in men and women separately, the waist circumference was found to be an independent factor in normal-weight men and women (data not shown).

In Asians, an impairment of early-phase insulin secretion plays an important role in the pathogenesis of type 2 diabetes,³⁸ and overall obesity is relatively uncommon. Asians have a higher proportion of visceral fat and a lower lean body mass than do white subjects with the same BMI.^15,39 Central obesity is the key factor that contributes to insulin resistance,⁵ and an increased visceral adiposity might be relevant in the pathogenesis of NAFLD.¹² Visceral adipose tissue is more resistant to insulin, exhibits a greater lipolysis, and produces more free fatty acid than does adipose tissue in other sites.⁴⁰ The increased availability of a substrate for lipogenesis and the relative hyperinsulinemia in insulin resistance status promote lipogenesis in the liver.⁴¹ The high prevalence of NAFLD in this study can be attributed to the relatively increased visceral fat depot in Asians, in addition to the high carbohydrate intake.

Our results found no difference between the normal-weight subjects with NAFLD and the overweight subjects without NAFLD in terms of the metabolic variables such as the waist-hip ratio, HOMA_IR, HOMA_{β-cell function}, and levels of fasting glucose, insulin, total cholesterol, HDL-C, and triglycerides. From the metabolic point of view, this finding suggests that normal-weight subjects with fatty liver are comparable to overweight subjects, although they have an apparently normal weight.

Moreover, in the normal-weight group, the odds ratios of insulin resistance and metabolic disorders such as hypertriglyceridemia, hyperuricemia, and central obesity in subjects with NAFLD compared with those without NAFLD were higher than the odds ratios in the overweight group. This suggests that NAFLD can be considered to be a more meaningful predictor of metabolic disorders, particularly in the normal-weight population.

The pathogenesis of NAFLD is still under active investigation. The “2-hits hypothesis” by Day and James⁴² suggests that the first hit involves the accumulation of excess fat in the hepatic parenchyma. This step has been linked to insulin resistance. The second hit is generally attributed to oxidative stress that causes the peroxidation of lipids in the hepatocyte membrane, which can initiate fibrosis via the proinflammatory cytokines and activated stellate cells. Lipid peroxidation and the generation of free radicals can also directly and adversely affect the hepatocytes, resulting in cell death and hepatic necrosis.

Considerable interest has been generated recently in nonalcoholic steatohepatitis, because the prevalence of cirrhosis was significantly higher among patients with nonalcoholic steatohepatitis than among patients with a steatosis alone.³ The limitation of this study was the lack of a histological confirmation of a diagnosis of nonalcoholic steatohepatitis.⁴³ Therefore, this study could not define the association between the histological significance and the metabolic abnormalities.

In summary, our study resulted in the following findings. First, NAFLD was a significant predictor of insulin resistance and metabolic disorders, and it may be anindicatorof metabolic disorders even in nonobese, nondiabetic people. Second, the significance of central obesity was more meaningful for NAFLD in the normal-weight group than in the overweight group, and this suggests that visceral adiposity rather than the overall amount of body fat is important, particularly in normal-weight people. Third, insulin resistance and central obesity were independently associated with NAFLD in normal-weight subjects, and this might have therapeutic implications. Insulin-sensitizing drugs or a reduction in central obesity through a nutritional regimen and exercise might reverse a fatty liver. Therefore, there is a need for further clinical studies. In conclusion, NAFLD is closely associated with metabolic disorders, regardless of obesity or diabetes, and can be considered an early predictor of metabolic disorders, particulary in the normal-weight population.

Correspondence: Bong Soo Cha, MD, PhD, Department of Internal Medicine, Yonsei University College of Medicine, 134 Shinchon-Dong, Seodaemoon-Ku, PO Box 120-749, Seoul, Korea (bscha@yumc.yonsei.ac.kr).

Accepted for Publication: November 28, 2003.

Financial Interest: None.

Funding/Support: This study was supported by grantR13-2002-054-01001-0 (2002) from the Basic Research Program of the Korean Science and Engineering Foundation, Daejeon.