NAFLD in children: A prospective clinical-pathological study and effect of lifestyle advice


  • Potential conflict of interest: Nothing to report.


Nonalcoholic fatty liver disease (NAFLD), a common cause of chronic liver disease in adults, is incompletely characterized in children. We conducted a prospective study to better characterize the clinical presentation of NAFLD in children and to determine the effect of lifestyle advice in the management of pediatric NAFLD. From June 2001 to April 2003, 84 children (age 3-18.8 yr) who had elevated aminotransferases and the diagnosis of NAFLD confirmed via liver biopsy underwent a 2-hour oral glucose tolerance test and a 12-month program of lifestyle advice consisting of diet and physical exercise. Thirty-four (40.5%) patients were obese (body mass index [BMI] >97th percentile), and 43 (51.2%) were overweight (BMI 85th-97th percentile). Ten (12%) had abnormal glucose tolerance; 10 (12%) had elevated triglycerides, cholesterol, or both; and all had normal blood pressure. Most children (67/84, 80%) were insulin-resistant, including the 7 children with normal BMI (<85th percentile). Increased liver fibrosis was present in 49 (58.1%) patients and was independently associated with obesity (OR 2.7, 95% CI 1.2-6.2) and age (1-year increase; OR 1.2, 95% CI 1.04-1.5). A 12-month program with diet and physical exercise resulted in a significant decrease in BMI, and levels of fasting glucose, insulin, lipids, and liver enzymes, as well as liver echogenicity on ultrasonography. In conclusion, children with NAFLD are almost always insulin-resistant regardless of BMI. Obesity and older age are independently associated with increased liver fibrosis. A simple lifestyle advice program significantly improves insulin resistance, and the liver disease in pediatric NAFLD. (HEPATOLOGY 2006;44:458–465.)

With the increasing prevalence of obesity and diabetes mellitus in the general population,1 nonalcoholic fatty liver disease (NAFLD) has become a common cause of chronic liver disease.2–5 NAFLD often clusters with conditions related to central obesity, including high triglyceride levels, low high-density lipoprotein cholesterol levels, high blood pressure, and hyperglycemia.6 NAFLD may present as a spectrum of disease from asymptomatic steatosis with or without elevated aminotransferases to cirrhosis with complications of liver failure and hepatocellular carcinoma.7 Recent studies on natural history have demonstrated that liver-related morbidity and mortality usually occur when the disease has progressed to advanced fibrosis and cirrhosis.8, 9

Over the last 4 decades, the prevalence of overweight and obesity in children has increased dramatically.10 In the United States, the prevalence of overweight and obesity among children aged 6 to 19 years was 31% and 16%, respectively, in 2001-2002 compared with a 5% obesity prevalence reported in 1963-1965.10 Given the strong association of NAFLD with increased body mass index (BMI), NAFLD affects a substantial proportion of children. The reported prevalence of NAFLD among children is 2.6%, but this increases to as much as 53% among obese children.11, 12 Paralleling the increasing prevalence of obesity in the pediatric population, NAFLD is expected to become one of the most common causes of end-stage liver disease in both children and young adults.

Although NAFLD is common and potentially serious, few series of pediatric NAFLD have been reported.13–18 Further data on the clinical presentation of NAFLD and the prevalence of underlying metabolic abnormalities in children are necessary. Additionally, there are few studies on the effect of therapeutic intervention with lifestyle modifications on liver disease in children. We report a prospective study of 84 children with biopsy-proven NAFLD. We sought to better characterize the presentation of NAFLD in children and to determine the beneficial effects of lifestyle intervention with diet and physical exercise in children with NAFLD.


NAFLD, nonalcoholic fatty liver disease; BMI, body mass index; OGTT, oral glucose tolerance test; HOMA, homeostatic model assessment; ISI, insulin sensitivity index; NAS, NAFLD activity score; NASH, nonalcoholic steatohepatitis; ALT, alanine aminotransferase.

Patients and Methods

The study met the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethics Committee of the Bambino Gesù Children's Hospital and Research Institute. Informed consent was obtained from all patients or their guardians.


Eighty-four consecutive, untreated patients (59 male, 25 female) with a mean age of 11.7 ± 3.3 years (range, 3-18.8) who were seen in our institution from June 2001 to April 2003 were included in the study. All patients were evaluated in our unit due to elevated serum aminotransferase levels. Aminotransferase levels were persistently elevated in 69 patients but fluctuated between normal and elevated in 15 patients. Patients with fluctuating levels had at least 2 measurements of elevated aminotransferases during medical follow-up before liver biopsy was performed. In all 84 patients, liver ultrasonography revealed increased echogenicity suggestive of steatosis, and liver biopsy confirmed NAFLD as the only explanation for the elevated aminotransferase levels. Polymerase chain reaction tests revealed that all patients were HCV RNA–negative. Secondary causes of steatosis, including alcohol abuse (≥140 g/wk), total parenteral nutrition, and the use of medications known to precipitate steatosis were excluded in all cases. Hepatitis A, B, C, D, E, and G; cytomegalovirus; and Epstein-Barr virus infections were ruled out with appropriate tests. In all cases, autoimmune liver disease, metabolic liver disease, Wilson's disease, and α-1-antitrypsin were ruled out using standard clinical and laboratory evaluation as well as liver biopsy features. To compare BMI among different ages and in both boys and girls, the BMI Z score was calculated. The Z score represents the number of standard deviations above or below the considered population mean value based on standardized tables for children.19 Obesity was defined as a BMI above 2 standard deviations that corresponds to a percentile above the 97th percentile adjusted for age and sex. Overweight was defined as a BMI within the 85th to the 97th percentile. Hypertriglyceridemia was defined as fasting triglyceride level >160 mg/dL, and hypercholesterolemia when fasting total cholesterol levels were >200 mg/dL.20–22

Evaluation of Glucose Metabolism and Insulin Sensitivity.

All patients underwent a 2-hour oral glucose tolerance test (OGTT) with the standard 1.75 g/kg of glucose, or a maximum of 75 g. Glucose tolerance status was determined according to the recently revised American Diabetes Association classification, in which fasting plasma glucose levels up to 99 mg/dL are considered normal; impaired fasting glucose is defined by a fasting plasma glucose level of 100 to 125 mg/dL; impaired glucose tolerance is defined by a 2-hour plasma glucose level of 140 to 199 mg/dL; and diabetes mellitus is defined by a fasting plasma glucose level ≥126 mg/dL or a 2-hour plasma glucose level ≥200 mg/dL.23

The degree of insulin sensitivity/resistance was determined via the homeostatic model assessment (HOMA) using the formula insulin resistance = (insulin × glucose)/22.524 and by the OGTT-derived insulin sensitivity index (ISI) using the formula ISI = (10,000/square root of (fasting glucose × fasting insulin) × (mean glucose × mean insulin during OGTT).25 Both the HOMA and the OGTT-derived ISI have a significant correlation with the gold standard euglycemic hyperinsulinemic glucose clamp technique. A HOMA value >2 or ISI value <6 were considered an indication of insulin resistance.

Liver Histology.

Biopsies were performed in all children using an automatic core biopsy device (Biopince, Amedic, Sweden) with a 150-mm-long 18-gauge needle able to cut tissue with extreme precision in lengths up to 33 mm.26 Liver biopsies were at least 15 mm long and were read by a single blinded liver pathologist. Biopsies were routinely processed (i.e., formalin-fixed and paraffin-embedded) and analyzed in sections stained with hematoxylin-eosin, Van Gieson, PAS-D, and Prussian blue stain. Iron storage was scored from 0 to 4 according to Searle et al.27 Immunohistochemical staining with α-1-antitrypsin was used to exclude α-1-antitrypsin deficiency–associated liver disease.

Histological features including steatosis, inflammation (portal and lobular), hepatocyte ballooning, and fibrosis were scored according to the scoring system for NAFLD recently developed by the NIH-sponsored NASH Clinical Research Network.28 The features of steatosis (0-3), lobular inflammation (0-3), and hepatocyte ballooning (0-2) were combined into a score that ranged from 0 to 8 known as the NAFLD activity score (NAS). Cases with NAS ≥5 have the diagnosis of nonalcoholic steatotohepatitis (NASH), cases with NAS ≤2 are considered “non-NASH,” and cases with a NAS of 3 or 4 are considered “borderline.”

Furthermore, the pattern of liver injury on liver biopsy was recorded to determine the proportion of children meeting criteria for type 1 and type 2 steatohepatitis as described recently by Schwimmer at al.18 Briefly, type 1 NASH refers to the presence of steatosis with hepatocyte ballooning and/or perisinusoidal fibrosis in the absence of portal features (the pattern of liver injury typically seen in adults). Type 2 NASH refers to the presence of steatosis along with portal inflammation and/or fibrosis in the absence of hepatocyte ballooning and perisinusoidal fibrosis.18

Lifestyle Advice Intervention.

All patients and guardians participated in a 1-hour nutritional counseling session by an experienced dietician. The patients were prescribed a balanced, low-calorie diet (25-30 cal/kg/d; carbohydrate, 50%-60%; fat, 23%-30%; protein, 15%-20%; fatty acid, two thirds saturated, one third unsaturated; ω6/ω3 ratio = 4:1) as recommended by the Italian Recommended Dietary Allowances. The diet was tailored to individual preferences. The goal of weight management was to attain a negative calorie balance. In addition to the prescribed diet, a moderate exercise program of aerobic exercise (30-45 min/d at least 3 times a week) was recommended and was tailored to individual preferences. Follow-up medical examinations and laboratory measurements were performed every 3 months during the first 12 months of the prescribed diet and physical exercise recommendations. Liver ultrasonography was repeated at 12 months. Compliance was monitored via monthly phone interviews and measurement of body weight and personal interview at the 3-month follow-up examinations.

Statistical Analysis.

Data are presented as the mean ± SD and number (proportion) of patients with a condition. Continuous variables were compared using nonparametric statistics as appropriate. Frequency data were compared using the chi-square test or Fisher exact test where appropriate. Spearman's rank correlation coefficient was used as a measure of association. Multivariate regression analysis was performed to look for independent predictors of liver fibrosis. A two-tailed P value of at least .05 was considered statistically significant.


Patient Characteristics.

The clinical details of the 84 children are summarized in Table 1. History of diabetes in first-degree relatives was not documented in any case. Insulin resistance as indicated by a HOMA score >2 or ISI score <6 was present in 52 (61.9%) and 62 (73.8%) children, respectively, with 67 (80%) of children having either a HOMA score >2 or an ISI score <6. Obesity was present in 34 (40.5%) children, 43 (51.2%) were overweight, and 7 had a BMI below the 85th percentile. Among these 3 BMI categories (obese, overweight, and normal BMI), HOMA values (2.49 ± 1.07 vs. 2.77 ± 1.45 vs. 3.18 ± 1.9, respectively; P = .6) and ISI values (4.88 ± 2.07 vs. 4.15 ± 1.93 vs. 3.45 ± 1.82, respectively; P = .1) were not significantly different. Also, the proportion of children with insulin resistance by either HOMA or ISI criteria was not significantly different between the 3 BMI categories (71%, 84%, and 100%, respectively; P = .2). Furthermore, BMI did not correlate significantly with HOMA (r2 = 0.01; P = .3), or ISI (r2 = 0.003; P = .6). Six (7.1%) children had hypertriglyceridemia, 9 (10.7%) had hypercholesterolemia, and 5 had both hypertriglyceridemia and hypercholesterolemia. Blood pressure was normal in all children.

Table 1. Anthropometric, Clinical, and Biochemical Characteristics of the Patients
VariablesMean ± SD or n (%)Range
  1. NOTE. Aminotransferase levels were persistently elevated in 69 patients but fluctuated between normal and elevated in 15 patients. Although aminotrasferase levels were within the normal range in these 15 patients at time of biopsy, each patient had at least 2 measurements of elevated aminostrasferases during the medical follow-up before the liver biopsy was performed. In all 84 children, liver echogenecity on ultrasonography suggested steatosis, and liver biopsy confirmed NAFLD as the only explanation for elevated aminostransferase levels.

  2. Abbreviations: BMI, body mass index; AST, aspartate aminotransferase; ALT, alanine aminotransferase; GGT, y-glutamyltranspeptidase; HOMA, homeostatic model assessment; ISI, insulin sensitivity index.

Age (yr)11.7 ± 3.33–18.8
Sex (M/F)59/25 
Pubertal stage (I, II, III, IV)29, 28, 14, 13 
BMI (kg/m2)26.3 ± 3.715.2–38.4
BMI Z score1.85 ± 0.7−1.72–5.1
Obesity (BMI >97th percentile)34 (40.5) 
Overweight (BMI 85th-97th percentile)43 (51.2) 
Normal (BMI <85th percentile)7 (8.3) 
Type II diabetes2 (2.3) 
Hypertension2 (2.3) 
AST (IU/L)48 ± 2619–187
ALT (IU/L)76 ± 6310–407
AST/ALT ratio0.8 ± 0.40.4–2.3
AST/ALT ration >116 (19) 
GGT (IU/L)26 ± 2110–130
Cholesterol (mg/dL)156 ± 3575–243
Hypercholesterolemia9 (10.7) 
Triglyceride (mg/dL)95 ± 5528–351
Hypertriglyceridemia6 (7.1) 
Fasting glucose (mg/dL)82 ± 1260–138
Fasting insulin (μU/L)11.8 ± 6.13.5–30.7
HOMA2.69 ± 1.350.7–6.7
HOMA >251 (60.7) 
ISI4.4 ± 2.01.2–9.1
ISI <662 (73.8) 

Glucose Metabolism.

Seventy-four (88%) children had normal glucose metabolism on OGTT, whereas the other 10 (12%) children (8 males, 2 female) showed impaired glucose metabolism, including 1 patient classified as having impaired fasting glucose, 7 patients classified as having impaired glucose tolerance, and 2 patients with previously unrecognized type 2 diabetes. These 10 patients were significantly older (13.7 ± 3.3 vs. 11.4 ± 3.3 yr; P < .05) and had significantly lower BMI Z scores (1.4 ± 1.2 vs. 1.9 ± 0.6; P = .05) than patients with normal glucose metabolism. No other clinical, laboratory, or liver histology features were statistically different between the 2 groups. However, the 10 (100%) children with abnormal glucose tolerance had insulin resistance by either HOMA or ISI parameters compared with 57 (77%) children with normal glucose metabolism (P = .09) Five out of 10 individuals with impaired glucose metabolism were in pubertal stage III (n = 1) or IV (n = 4); the other 5 children were in pubertal stage I (n = 1) or II (n = 4).

Liver Histology.

The histological findings are summarized in Table 2. All biopsies showed steatosis to be mostly macrovesicular but frequently associated with microvesicular steatosis in some hepatocytes. The pattern of steatosis was diffuse or scattered lobular, and showed zonal distribution in only 10 cases. The inflammatory infiltrate was mainly composed of lymphocytes and neutrophils; when granulomas were present, mononuclear histiocytic cells and eosinophils were also present. Apoptotic cells were noted occasionally. Glycogenated nuclei of variable dimensions were present in 46 of the 84 cases; this nuclear change was noted mostly in zone 1. No Mallory hyaline was noted in any case, and mild iron deposition was present in 3 cases.

Table 2. Histological Findings of the Patients
Grade/StageSteatosis, n (%)Necroinflammation, n (%)Ballooning, (n %)Fibrosis, (n %)
  1. NOTE. Biopsies were scored according to Kleiner et al.28

013 (14.4%)43 (51.2%)35 (41.6%)
130 (35.7%)63 (75%)23 (27.4%)43 (51.1%)
234 (40.4%)8 (9.5%)18 (21.4%)2 (2.3%)
320 (23.8%)4 (4.7%)

The average NAS28 was 3.5 ± 1.5 and ranged from 1 to 7. According to the NAS, 22 (26.2%) patients met criteria for the diagnosis of NASH (NAS ≥5), and 28 (33.3%) patients were labeled as non-NASH (NAS ≤2); the remaining 34 (40.5%) patients were considered borderline (NAS 3-4). When the Schwimmer et al.18 criteria were applied, only 2 (2.4%) patients met the criteria for NASH type 1, and 24 (28.6%) patients met the criteria for NASH type 2. There was overlap of NASH type 1 and 2 in 44 (52.4%) patients, and the remaining 14 (16.7%) patients had simple steatosis on liver biopsy.

Liver Fibrosis.

Increased fibrosis was noted in 49 (58%) patients but was mostly of mild (stage 1) severity, with only 4 (4.7%) patients showing septal fibrosis (stage 3). Among the 43 patients with stage 1 fibrosis, 3 were 1a, 7 were 1b, and 33 were 1c. None of the patients showed cirrhosis-stage disease on liver biopsy. Table 3 shows the comparison between patients with and without liver fibrosis. Patients with liver fibrosis were significantly older, had significantly higher BMI, and had higher serum levels of cholesterol and triglycerides compared with those without fibrosis. In a multivariate logistic analysis, however, only age and BMI remained statistically significant. As summarized in Table 4, every 1-year increase in age was associated with a 1.2-fold (CI 1.04-1.5) increased risk of having liver fibrosis. Compared with patients with normal BMI, those with obesity had a 2.7-fold (95% CI 1.2-6.2) increased risk of having liver fibrosis, whereas overweight patients were not at a higher risk (OR 0.9, 95% CI 0.4-1.9). Patients with liver fibrosis had significantly higher inflammation and hepatocyte ballooning scores, and consequently a significantly higher NAFLD activity score compared with patients without liver fibrosis (Table 3).

Table 3. Anthropometric, Clinical, and Biochemical Characteristics of Patients With and Without Liver Fibrosis
VariablesWith Liver Fibrosis (n = 49)Without Liver Fibrosis (n = 35)P Value
  1. Abbreviations: BMI, body mass index; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; AST, aspartate aminotransferase; ALT, alanine aminotransferase; γGT, gamma-glutamyltranspeptidase; HOMA, homeostatic model assessment method; ISI, insulin sensitivity index; NAFLD, nonalcoholic fatty liver disease.

Age (yr), mean ± SD12.3 ± 3.310.9 ± 3.3.05
Sex, M/F5/1424/11.8
BMI (kg/m2), mean ± SD27.3 ± 3.825 ± 3.1.004
BMI Z score, mean ± SD2.0 ± 0.71.7 ± 0.8.05
BMI category  .08
Obesity (BMI >97th percentile), n (%)24 (49%)10 (28.6%) 
Overweight (BMI 85th-97th percentile), n (%)23 (46.9%)20 (57.2%) 
Normal (BMI <85th percentile), n (%)  (4.1%)5 (14.3%) 
Diabetes, IFG, IGT (n)2, 0, 50, 1, 2.3
Hypertension, n (%)  (4%)0.2
AST (IU/L), mean ± SD52 ± 3144 ± 15.4
ALT (IU/L), mean ± SD81 ± 7968 ± 27.7
AST/ALT ratio, mean ± SD0.86 ± 0.50.71 ± 0.3.5
AST/ALT >1, n (%)11 (22.4%)5 (14.3%).4
GGT (IU/L), mean ± SD26 ± 1925 ± 24.5
Cholesterol (mg/dL), mean ± SD163 ± 34145 ± 33.01
Hypercholeterolemia, n (%)6 (12.2%)3 (8.6%).7
Triglyceride (mg/dL), mean ± SD106 ± 5480 ± 54.002
Hypertriglyceridemia, n (%) 5 (10.2%)1 (2.9%).4
Fasting glucose (mg/dL), mean ± SD82 ± 1182 ± 9.9
Fasting insulin (μU/L), mean ± SD11.4 ± 5.512.4 ± 6.9.8
HOMA, mean ± SD2.4 ± 1.13.1 ± 1.6.06
ISI, mean ± SD4.7 ± 2.04.0 ± 2.0.1
Steatosis score, mean ± SD2.1 ± 0.71.7 ± 0.8.02
Lobular inflammation score, mean ± SD1.1 ± 0.40.7 ± 0.5<.001
Hepatocyte ballooning score, mean ± SD1.1 ± 0.80.1 ± 0.6<.001
NAFLD activty score, mean ± SD4.3 ± 1.42.5 ± 1.0<.001
Table 4. Multivariate Logistic Regression Analysis of Association With Presence of Liver Fibrosis
VariableOR95% CIP Value
Age (continuous)<.02
BMI category    
Normal (BMI <85th percentile)Reference
Overweight (BMI 85th-97th percentile)
Obesity (BMI >97th percentile)
Cholesterol (continuous)1.30.981.6.2
Triglycerides (continuous)

Effect of Lifestyle Advice.

Fifty-seven of the 84 patients had completed 1 year of lifestyle advice consisting of diet and increased physical exercise. The comparison between the baseline and 1-year data is summarized in Table 5. The decrease in BMI was associated with a significant improvement in liver enzyme, triglyceride, cholesterol, insulin, and glucose levels. In addition, weight loss led to a significant improvement in insulin sensitivity as indicated by a significant decrease in HOMA. The temporal changes of aminotransferase levels, body weight, and BMI are illustrated in Fig. 1.

Table 5. Comparison of Baseline Values and Those Obtained at 1 Year of Lifestyle Advice
 Baseline (n = 57)At 1 Year (n = 57)P Value
  1. NOTE. All values other than P values are expressed as the mean ± SD.

  2. Abbreviations: BMI, body mass index; AST, aspartate aminotransferase; ALT, alanine aminotransferase; GGT, gamma-glutamyltranspeptidase; HOMA, homeostatic model assessment method.

BMI (kg/m2)25.9 ± 3.623.8 ± 2.7<.0001
Weight (kg)60.9 ± 17.156 ± 14.7<.0001
AST (IU/L)43 ± 1631 ± 9<.0001
ALT (IU/L)62 ± 3133 ± 10<.0001
GGT (IU/L)24 ± 2021 ±11.02
Cholesterol (mg/dL)152 ± 33128 ± 17<.0001
Triglyceride (mg/dL)85 ± 4666 ± 30<.0001
Fasting glucose (mg/dL)82 ± 1271 ± 7<.0001
Fasting insulin (μU/L)12.3 ± 6.510.2 ± 5<.0001
HOMA2.54 ± 1.41.71 ± 0.9<.0001
Figure 1.

Changes in serum aminotransferase levels, BMI, and body weight during a 12-month period of lifestyle advice with a balanced, low-calorie diet and increased physical activity. Columns represent mean values, and error bars represent standard deviation. The changes from baseline to 12 months were statistically significant (P < .0001 for all comparisons). The normal range for aminotransferase levels is 5 to 40 IU/L. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index.

Five of these 57 patients had a BMI below the 85th percentile at baseline. After 12 months of lifestyle advice, body weight remained essentially unchanged (−2.5 ± 1.4%), but alanine aminotransferase (ALT) levels improved in 2 patients and normalized in 3 patients. Liver echogenicity suggested complete resolution of steatosis in 1 patient, improvement in 3 patients, and no change in 1 patient whose ALT level had normalized (from 89 IU/L to 24 IU/L).

Fifty-two of these 57 patients had a BMI at the 85th percentile or higher at baseline. After 12 months of lifestyle advice, their body weight decreased by 8 ± 4.7%. Body weight decreased by more than 10% of baseline in 17 patients (with liver echogenicity suggesting complete resolution of steatosis in 5 patients and improvement in 12 patients), from 5% to 10% in 17 patients (with liver echogenicity suggesting improvement of steatosis in all), and to <5% in 18 patients (with liver echogenicity suggesting improvement of steatosis in 12 patients and no change in 6 patients). The 6 patients with weight loss <5% and no changes in liver echogenicity still had improvement in ALT levels (4 of these patients reached the normal level). Overall, patients losing 5% or more of body weight had higher improvements in ALT levels compared with those with <5% weight loss (−35 ± 33 vs. −20 ± 20 IU/L, respectively; P < .05). Among these 52 patients, 11 had ALT levels above the upper limit of normal (40 IU/L) at 12 months of lifestyle advice, with a similar proportion between those losing ≥5% or <5% of body weight (6/34 [18%] vs. 5/18 [28%], respectively; P = .4). However, ALT levels had improved in 9 of these 11 children at 12 months of lifestyle advice.


This large prospective study of children with liver biopsy–proven NAFLD provides important new data and extends prior observations on the clinical presentation and the effect of diet and increased physical activity on NAFLD in the pediatric population. Our study demonstrates that: (1) NAFLD is common in children with elevated aminotransferases, with most patients being either obese or overweight; (2) most children with NAFLD are insulin-resistant despite having normal OGTT and normal BMI; (3) more than half of children with NAFLD undergoing liver biopsy have some degree of liver fibrosis; (4) older age and presence of obesity significantly—and independently of other risk factors—increase the odds of having liver fibrosis; and (5) lifestyle advice with diet and increased physical activity significantly improves insulin sensitivity, liver enzymes, and liver echogenicity on ultrasonography in children with NAFLD.

NALFD is considered the hepatic manifestation of the metabolic syndrome, a cluster of abnormalities related to insulin resistance, of which obesity is one of the major components. Although a clinical definition of the metabolic syndrome in children does not currently exist, its prevalence based on the criteria proposed by the Adult Treatment Panel III of the National Education Program was 6.7% in adults aged 20-29 years and 4.2% in adolescents.29 However, the perspective is very different in overweight/obese adolescents. Using data from the National Health and Nutrition Examination Survey III (1988-1994), the reported prevalence of the metabolic syndrome in adolescents with a BMI at the 95th percentile or higher in the United States is 28.7% compared with 6.1% in adolescents with a BMI between the 85th percentile and the 95th percentile, and 0.1% in those with a BMI below the 85th percentile.29 Thus, the prevalence of the metabolic syndrome in children is clearly related to both increasing BMI and older age. Consistent with this observation, 77 (91.7%) of our patients with NAFLD were either obese or overweight. The prevalence of obesity as defined by a BMI above the 97th percentile was 40.5% in our series, or 68% (57/84) if the 95th percentile or higher was used. Ethnic differences, along with the prospective design of our study, may explain this lower prevalence of obesity in our series compared with others.13–18

Although insulin resistance using HOMA or ISI criteria was present in most of our patients with NAFLD (67/84 [80%]), 17 (20%) patients did not meet criteria for insulin resistance. No clinical, laboratory (other than insulin resistance), or histological features were significantly different between these 2 groups (data not shown). Interestingly, BMI did not significantly correlate with presence of insulin resistance; the 7 patients with normal BMI had an ISI score <6, and 6 of them had a HOMA score >2. Thus, NAFLD in children is associated with insulin resistance despite normal BMI. This finding is consistent with studies in adults demonstrating an association of insulin resistance and NAFLD regardless of BMI.30 We did not measure body fat distribution in our patients; thus, it remains unclear whether the observation in adults that adiposity with a central distribution is more often associated with NAFLD regardless of BMI31 also applies to children. Interestingly, the 7 patients in our study with normal BMI also had normal levels of total cholesterol and triglycerides, with 6 of these patients having normal OGTT results (1 had impaired fasting glucose). This indicates that some children with NAFLD with a normal BMI, normal glucose tolerance, and a normal lipid profile may still meet criteria for insulin resistance, and that insulin resistance and NAFLD may precede the development of obesity, overt diabetes, and dyslipidemia later in life.

We applied the 2 scoring systems recently proposed for grading and staging the histological features of NAFLD in children.18, 28 Using the criteria of Kleiner et al.,28 which were developed for use in clinical trials, the prevalence of NASH, borderline NASH, and non-NASH was 26.2%, 40.5%, and 33.3%, respectively. However, there is no consensus on the most appropriate definition of NASH. If NASH is indeed the progressive form of NAFLD, and therapy should be targeted toward those with NASH to prevent disease progression, then we believe any patients with NAFLD and increased fibrosis should be classified as having NASH and should be offered enrollment in clinical trials regardless of presence and severity of inflammation or hepatocyte ballooning. Presence of fibrosis may be a better indicator of a worse prognosis compared with other features of NASH such as steatosis, hepatocyte ballooning, and inflammation; these other features tend to improve or disappear as fibrosis progresses over time.32 Twenty-five of the 34 patients categorized as borderline in our series had increased liver fibrosis.

Using the criteria of Schwimmer et al.,18 the prevalence of simple steatosis, NASH type 1, NASH type 2, and overlap of NASH type 1/2 in our patients was 16.7%, 2.4%, 28.6%, and 52.4%, respectively, compared with 16%, 17%, 51%, and 16% in the Schwimmer et al.18 study. Thus, NASH type 1 is rarely seen in children. However, unlike the Schwimmer et al.18 series, type 2 NASH was not the prevalent form of liver injury in our children, but an overlap of type 1 and 2 pattern of liver injury. Most of our children (69.1%) had portal injury placing them as either NASH type 2 or overlap of type 1 and 2 NASH. Further studies are necessary to validate these results and to develop a more reproducible scoring system to interpret liver histology in pediatric NAFLD.

More than half (58.4%) of our patients had increased liver fibrosis, including some with septal fibrosis. Given the prospective design of our study, we believe this figure gives a better estimate of the prevalence of fibrosis in children with NAFLD compared with retrospective series. Presence of fibrosis on liver biopsy is a worrisome finding, because it suggests more severe and progressive liver injury. As has been previously found in adults,33, 34 older age and presence of obesity were the only independent indicators of presence of liver fibrosis in our series. Increased BMI was identified as an indicator of perisinusoidal fibrosis in a series of 43 children with NAFLD.17 This finding suggests that in children with NAFLD, the risk of developing liver fibrosis and progressing to a more advanced stage of disease is time-dependent and that the presence of obesity increases the risk of fibrosis development.

Our study shows that a simple lifestyle advice program consisting of diet and physical exercise led to a significant improvement of insulin sensitivity, lipid levels, and liver enzymes. Changes in lifestyle comprise a basic approach to the treatment of the components of the metabolic syndrome and should constitute the first step in the management of children with NAFLD. Increased physical activity per se improves insulin sensitivity in muscle35–37 and reduces visceral fat regardless of weight loss.38 In fact, the 5 patients with normal BMI in our series showed improvement or normalization of aminotransferase levels, with improvement of liver echogenicity on ultrasonography in 4 patients, despite trivial or no change in body weight. Thus, lifestyle advice may be beneficial even in those children who are not obese or overweight.

Among overweight/obese children, those losing at least 5% of body weight showed a greater decrease in ALT levels; however, improvement or normalization of ALT was also seen among those losing less than 5% of body weight including improvement of echogenicity on ultrasonography in some of them. Thus, even mild weight loss in overweight/obese children with NAFLD may improve insulin sensitivity and liver enzymes. A limitation of our study is that we did not perform posttreatment liver biopsy in our patients; thus, further studies correlating changes in liver enzymes and liver histology are necessary. In addition, the lifestyle advice in our series was not a randomized, double-blinded, controlled trial; furthermore, it remains unclear whether the addition of medications that increase insulin sensitivity or the antioxidant defenses in the liver increase the efficacy of lifestyle intervention.

In summary, in this large prospective series, NAFLD was a common cause of abnormal liver enzymes in children with increased BMI, but it also occurred in children with normal BMI and absence of other components of the metabolic syndrome. More than half of our children had increased liver fibrosis, which was associated with older age and presence of obesity. The vast majority of children with NAFLD had underlying insulin resistance, and insulin resistance, liver enzymes, and liver echogenicity on ultrasonography improved significantly with lifestyle advice consisting of diet and increased physical activity. This study contributes significantly to our understanding of the clinical presentation of NAFLD in the pediatric population and highlights the potential benefit of lifestyle advice as a useful treatment modality.