Dr V. Nobili, Liver Unit, Research Institute, Bambino Gesu’ Children's Hospital, S. Onofrio 4 Square, 00165 Rome, Italy. E-mail: firstname.lastname@example.org
Background Few data are available on the effect of antioxidants in paediatric non-alcoholic fatty liver disease (NAFLD).
Aim To compare the effect of a nutritional programme alone or combined with alpha-tocopherol and ascorbic acid on alanine aminotransferase (ALT) levels, and insulin resistance (IR) in biopsy-proven NAFLD children.
Methods In a 12-month double-blind placebo study, 90 patients were prescribed a balanced calorie diet (25–30 cal/kg/d), physical exercise, and placebo (group A) or alpha-tocopherol 600 IU/day plus ascorbic acid 500 mg/day (group B). IR was estimated by the homeostasis model assessment (HOMA-IR).
Results At month 12, ALT (32.67 ± 8.09 vs. 32.18 ± 11.39 IU/L; P = NS), HOMA-IR (1.52 ± 0.66 vs. 1.84 ± 0.95 IU/L; P = NS), and weight loss (32% vs. 35% of excessive body weight; P = NS) did not differ between the two arms. Among subjects who lost ≥20% of their excessive weight, ALT and body weight percentage changes were significantly related (ro = 0.260; P = 0.03). In subjects, who lost more than 1.0 kg, HOMA-IR significantly decreased (2.20 ± 0.21 to 1.57 ± 0.13 in group A (P ≤ 0.01; −8%); 2.91 ± 0.24 to 1.88 ± 0.16 in group B (−32%; P ≤ 0.0001)). ALT decreased by 36% (59.13 ± 4.11 vs. 30.27 ± 1.46 IU/L; P ≤ 0.001), and 42% (68.19 ± 5.68 vs. 31.92 ± 1.92 IU/L; P ≤ 0.0001). In a multivariate analysis, fasting insulin changes in group A (P = 0.012; F = 7.150).
Conclusions Diet and physical exercise in NAFLD children seem to lead to a significant improvement of liver function and glucose metabolism beyond any antioxidant therapy.
Non-alcoholic fatty liver disease (NAFLD) is increasingly recognized as a common cause of chronic liver disease in children. This condition is frequently associated with obesity. In Italy, one-fourth of obese children have increased levels of serum aminotransferases,1 and, conversely, 40% of biopsy-proven NAFLD children is obese.2 Insulin resistance (IR) plays a pivotal role in NAFLD.3
In childhood, so far, low calorie diet2, 4, 5 and life-style advice2 are the only established therapeutic options. Lack of adherence to dieting programmes limits the utility of these treatments. Hence, additional therapeutic options are needed.
Based on the hypothesis that injury from oxidative stress contributes to both the progression from NAFLD to overt non-alcoholic steatohepatitis (NASH)6 and the worsening of IR,7 antioxidative treatments aimed at decreasing oxidative stress have been conducted in NAFLD children8, 9 and adults10–12 with not conclusive results.
In liver, chronic exposure to oxidative stress, because of the oxidation of cytotoxic free fatty acids (FFA), can up-regulate cytokines production,13 induce cytochrome P450 enzyme 2E1 expression, and significantly reduce liver concentration of antioxidants.14 Enhanced cytokine production triggers hepatic stellate cell activation,15 leading to fibrogenesis and extra-cellular matrix protein deposition. In in vitro studies,16, 17 vitamin E inhibits pro-inflammatory cytokine production and attenuates hepatic fibrosis/collagen production. Its supplementation significantly restores hepatic glutathione reserves, reducing oxidative stress, hepatic fibrosis18 and collagen deposition19 in murine models of NASH.
In muscle tissue, the main site of oxidative processes, FFA overflow can overload the mitochondrial oxidation process, leading to accelerated production of reactive oxygen species (ROS),20 which interfere with insulin signalling.21 High-dose vitamin E supplementation may improve insulin action and decrease plasma fasting insulin and glucose levels by decreasing cellular oxidant stress, altering membrane properties,7, 22, 23 and decreasing inflammatory activity. Increased vitamin E may enhance the endogenous cellular antioxidant defence system and reduce levels of ROS that are produced by mitochondria. Vitamin E can also act at the cellular level independently of its antioxidant activity and may potentially contribute to improved insulin action through the inhibition of protein kinase C;24 the decrease of intracellular levels of diacylglycerol25 and the activation of insulin substrate protein-1.26
It is conceivable that circulating levels of vitamin E, being liposoluble, may result to be low in obese subjects.27 A diet low in vitamins and rich in fats may further contribute to deplete vitamin E depots. Thus, insufficient quantities of circulating antioxidants cannot counteract ROS in obese subjects prone to both NAFLD and IR.
To the best of our knowledge, two clinical studies have been performed in NAFLD children with the aim of testing the effect of vitamin E on liver laboratory tests and brightness.8, 9 No one looked at the relation among vitamin E supplementation, liver function and IR if there is any. In the present double-blind 12-month study, we compared the effect of vitamin E (600 IU/day) vs. placebo on liver function, brightness and IR in 90 biopsy-proven NAFLD children undergoing a nutritional programme of a balanced calorie diet plus daily physical exercise. Vitamin C (500 mg/day) was also administered as it enhances regeneration of oxidized vitamin E.28
Patient selection and evaluation
Ninety consecutive otherwise healthy biopsy-proven NAFLD children (aged 3–18 years) were enrolled from January 2003 to March 2005. Inclusion criteria were persistently elevated serum aminotransferase levels, diffusely echogenic liver in imaging studies suggestive of fatty liver, exclusion of hepatic virus infections (HCV-RNA PCR negative, hepatitis A, B, C, D, E and G, cytomegalovirus and Epstein–Barr virus) no alcohol consumption, no history of parenteral nutrition, no use of drugs known to induce steatosis (e.g. valproate, amiodarone or prednisone). All children had negative serologic markers for known chronic liver disease. Presence of antibodies (antinuclear, smooth muscle, and antiliver-kidney microsomal antibodies) was tested. Levels of circulating immunoglobulins were quantified. Ceruloplasmin and alpha-1-antitrypsin levels were measured. Expression of alpha-1-antitrypsin phenotype was also investigated. Circulating serum iron and iron-binding capacity were estimated. Cholesterol ester storage disease, cystic fibrosis, tyrosinemia type 1, homocystinuria, glycogen-storage diseases were rule out by appropriate tests.
The body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in metres. To compare BMI across different ages and in both boys and girls, BMI Z-score was considered. The Z-score represents the number of s.d. above or below the considered population mean value based on standardized tables for children.29 Obesity was defined as a BMI above 2 s.d. which correspond to the 97th adjusted for age and gender. Using accepted conventions, we defined ‘overweight’ as those children with a BMI between the 85th and 96th percentile for age and gender.30, 31 Standard growth charts were used to determine per cent of ideal body weight (IBW) at the baseline as median weight (50th) for length from sex-specified charts. Excessive body weight (EBW) was computed as body weight minus IBW.
Insulin resistance was determined by the homeostasis model assessment (HOMA-IR) using the formula: IR = (insulin×glucose)/22.5.32
All children underwent liver biopsy. Biopsies were routinely processed and analysed.2 The main histological features of NAFLD/NASH, which include steatosis, inflammation (portal and lobular), hepatocyte ballooning, and fibrosis were scored by using the scoring system for NAFLD according to the guidelines of the NIH-sponsored NASH Clinical Research Network as elsewhere described.33
Plasma glucose was measured in triplicate by the glucose oxidase technique on a Beckman glucose analyzer (Beckman, Fullerton, CA, USA); plasma insulin by a specific radio-immuno-assay (MYRIA Technogenetics, Milan, Italy). Serum triglycerides, and total and high density lipoprotein (HDL) cholesterol were measured spectrophotometrically.
Written informed consent was obtained by parents. The study was approved by the Ethical Committee at the Bambino Gesù Hospital.
Patients were randomly assigned to a 12-month double-blind treatment of nutritional counselling plus placebo (group A) or diet plus oral vitamin supplementation [vitamin E 600 IU/day and vitamin C 500 mg/day (group B)]. Nutritional counselling consisted of a low-calorie diet for subjects with a BMI ≥ 85th percentile and an iso-calorie diet for normal weight subjects. Patients and responsible guardians underwent 1-hour nutritional counselling monthly by the same experienced dietician. Diet was tailored on individual preferences and balanced as recommended by the Italian Recommended Dietary Allowances. (hypo-caloric diet 25–30 cal/kg/day, iso-caloric diet 40–45 cal/kg/day, carbohydrate 50–60%; fat 23–30%; protein 15–20%; fatty acid: two-third saturated, one-third unsaturated; ω6/ω3 ratio = 4:1). The goal of weight management was to induce a negative calorie balance in subjects with a BMI≥ 85th percentile and to allow a balanced diet in normal weight subjects. The above diet regimen was associated with a recommendation to engage in a moderate daily exercise programme (45 min/day aerobic physical exercise). Vitamins and placebo were prepared by the same chemist at Bambino Gesù Hospital. The treatment period lasted for 12 months.
Laboratory tests were performed at the baseline and every 3 months. Blinded ultrasonographic evaluation of liver brightness was assessed by the same physician at the baseline and after 12-month treatment.
In both groups compliance was tested by pill counting and amount of weight loss. Parents were recommended to personally verify the daily intake of medications. Compliance was considered to be good when patients took more than 90% of the pills provided, as verified by counting residual pills at the next visit to the out-patient department.
Statistical analysis took into account compliance to both drug and diet treatments. In those patients with a BMI ≥ 85th percentile, a weight loss of at least ≥20% of the initial EBW or a decrease in body weight ≥1.0 kg was suggestive of a good compliance to the prescribed diet.
Treatments were considered a priori equivalent, and the analysis was planned to compare vitamin treated vs. only dieting subjects. The primary outcome was ALT normalization and patients whose ALT levels were reduced below the upper normal limit were considered as ALT responders at month 12. We considered an average 35% ALT normalization by the diet approach. Considering a type I error of 0.05 and a type II error of 0.20, 43 subjects for arm were needed to achieve statistical significance.
Data were presented as mean±s.d. Continuous variables were compared using the non-parametric Mann–Whitney U-test for unpaired data, whereas the Wilcoxon signed ranks test was used for comparison of paired data. Frequency data were compared using chi-squared test or Fisher's exact test where appropriate. To compare variable treatment-induced changes attributable to weight loss only or to vitamin supplementation, it was necessary to carry out an additional post hoc analysis in subgroup of patients selected on the basis of amount of weight loss alone and to the vitamin supplementation. Relationships between variables were sought by linear correlation analysis (Spearman's r) and multivariate analysis performed using standard techniques. A two-tailed P-value <0.05 was considered statistically significant. spss 11.5 for Windows (SPSS Inc., Chicago, IL, USA) was used for statistical analysis.
Baseline characteristics and liver histology
Table 1 lists baseline and 12-month characteristics of the two groups. Two patients from the placebo group dropped out their study. The two arms were balanced for all parameters, including histology pattern, but except fasting glucose (P = 0.001) and HOMA-IR (P = 0.03) at the baseline. In group A, 10 patients were normal weight, 17 had a BMI between the 85th and 96th percentile, and in 16 subjects overt obesity was diagnosed. In group B, eight children were normal weight, 19 at risk for obesity or otherwise overweight, and 18 obese (Figure 1).
Table 1. Clinical and laboratory characteristics of the two groups at the enrolment and at the end of the study
Group A (diet plus placebo) n = 43
Group B (diet plus vit E 600 IU/day plus vit C 500 mg/day) n = 45
At 12 months
At 12 months
Data are expressed as mean ± s.d. Significance at the intra-group comparison is reported (Wilcoxon test). In group A, two patients were lost to the follow-up and not included in the analysis.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; γ-GT, gamma-glutamyl transpeptidase; BMI, body mass index, HOMA-IR, homeostasis model assessment.
* anova followed by post hoc analysis showed significant differences in fasting glucose between group A and B either before (P = 0.001) or after treatment (P = 0.05), and in HOMA-IR values at the baseline (P = 0.05).
12.36 ± 3.02
12.07 ± 3.29
25.46 ± 3.60
23.63 ± 2.77
26.49 ± 3.57
24.10 ± 2.62
59.84 ± 17.79
55.37 ± 15.47
64.50 ± 16.47
58.78 ± 14.26
57.11 ± 23.63
32.67 ± 8.09
68.53 ± 33.34
32.18 ± 11.39
41.22 ± 12.30
31.17 ± 7.69
45.21 ± 17.28
31.26 ± 8.96
21.47 ± 8.41
20.42 ± 7.35
27 ± 9.60
22 ± 14.62
Fasting glucose (mg/dL) *
77.30 ± 9.19
69.23 ± 2.92
86.03 ± 12.47
72.44 ± 7.99
Fasting insulin (μIU/mL)
11.05 ± 5.34
9.58 ± 4.72
13.14 ± 6.72
10.56 ± 4.83
2.11 ± 1.09
1.52 ± 0.66
2.82 ± 1.48
1.84 ± 0.95
154.39 ± 37.39
129.61 ± 17.28
152.76 ± 34.18
127.08 ± 19.09
87.53 ± 52.01
66.72 ± 32.11
80.24 ± 3.08
64.31 ± 26.66
Steatosis was present in all biopsies, mostly macrovescicular, but frequently associated with microvesicular steatosis. The pattern of steatosis was diffuse or scattered lobular. In 10 cases, it was zonally distributed. In 77 children, inflammation was present. Hepatocyte ballooning was present in 46 patients. Glycogenated nuclei of variable dimension were present in 50 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 four cases. The average NAFLD activity score (NAS) was 3.5 ± 1.5 and ranged from 1 to 7. According to the NAS, 24 patients met criteria for the diagnosis of NASH (NAS ≥ 5), and 31 patients were labelled as non-NASH (NAS ≤ 2); the remaining 35 patients were considered borderline (NAS 3–4). Increased fibrosis was noted in 54 but mostly of mild (stage 1) severity with only five children showing septal fibrosis (stage 3). No patient showed cirrhotic stage disease on liver biopsy.
Characteristics at the end of the study period
At the end of the study, levels of fasting glucose were still significantly different between the two groups, being higher in the vitamin treated group (P = 0.05), while ALT levels, HOMA-IR, amount of weight loss did not differ (Table 1). All enrolled subjects were compliant to placebo or vitamin administration. In fact, compliance, estimated as percentage of pills taken during the treatment, was equally larger than 90% in each subject of both group (mean values of 93% vs. 94%; P = NS). As far as compliance to the diet treatment concerns, in group A, 29 of 33 obese and overweight patients lost more than 20% of their EBW, and 37 of 37 patients in group B (P = NS). Patients lost an average of 32.7% in the group A and 35.73% of their EBW in the group B (P = NS). A weight loss ≥1.0 kg was respectively observed in all 33 children from group A, who needed to lose weight and 35 of 37 from the vitamin treated arm (P = NS).
Normalization of ALT levels were observed in 26 patients (31%) in the placebo group and in 32 patient in the vitamin treated arm (38%; P = NS). In the placebo group, liver brightness was unchanged in seven patients, reduced in 37; and it disappeared in three children. In the vitamin treated group liver brightness was unchanged in five patients, reduced in 33; and it disappeared in three children.
Stratification of the two groups of patients on the basis of the effective weight loss defined as a weight loss ≥20% of EBW, and compliance to vitamin therapy (≥90%) yielded four subgroups (Figure 1): in the placebo arm, weight (Group A1) and not weight loser patients (group A2); in the vitamin treated arm, weight loser patients compliers to vitamin treatment (Group B1), and not weight loser children, but compliers to the antioxidant therapy (Group B2). After reallocating patients in the four groups, significant statistical differences were found in weight loss, HOMA-IR, ALT levels, fasting insulin and triglycerides at the intra-group comparisons (Figure 2, data not reported, P-values ranging from 0.03 to 0.0001), but no significant difference was found at the inter-group comparison (P = NS). Among those subjects who lost ≥20% of their excessive weight, percentage of changes in ALT levels and in body weight were significantly related (ro = 0.260; P = 0.03). Figure 3 shows changes in body weight and ALT levels over the treatment period.
Groups A and B were stratified according to a weight loss ≥1.0 kg (Figure 1). In subjects, who lost more than 1.0 kg, HOMA-IR decreased from 2.20 ± 0.21 to 1.57 ± 0.13 in group A (P ≤ 0.01; −13.57 ± 12.98%) and from 2.91 ± 0.24 to 1.88 ± 0.16 in group B (P ≤ 0.0001; −32.30 ± 17.98%). ALT levels also decreased (−36.38 ± 27.50%; 59.13 ± 4.11 vs. 30.27 ± 1.46 IU/L; P ≤ 0.001 in group A; and 42.64 ± 29.09%; 68.19 ± 5.68 vs. 31.92 ± 1.92 IU/L; P ≤ 0.0001 in group B). In subjects who did not lose weight, HOMA-IR decreased by 19.49% (±17.81; from 1.68 ± 0.16 to 1.28 ± 0.05; P ≤ 0.01) in the placebo group, and not significantly in the vitamin treated arm (1.13 ± 0.26 vs. 1.12 ± 0.26; P = NS; −21.91 ± 20.05%). ALT levels improved by 8.44% (±31.74; 47 ± 11.66 vs. 35.17 ± 4.17 IU/L; P = 0.05) and 39.55% (±25.53%; 74.50 ± 14.50–37 ± 8 IU/L; P ≤ 0.01), respectively.
In a multivariate analysis, changes in HOMA-IR and ALT levels were put as dependent variables, whilst changes in body weight, BMI, fasting insulin and triglycerides were independent variables. In the placebo group, changes in fasting insulin were found to be a predictor of changes in ALT levels (P = 0.012; F =7.150) after treatment. The model explained 47% of the variance.
Since the presence of 10 normal weight subjects in group A2 and eight in group B2, we looked at the effect of vitamin treatment in these subjects. The lack of any significant differences in body weight, ALT levels, HOMA-IR, fasting glucose and insulin (data not reported) between these two arms was once more confirmed.
Side effects or adverse events
No side effects or adverse events were recorded.
Addition of antioxidant vitamins, vitamin E and C to a healthier diet and daily regular physical activity seems to not add any significant effect in terms of reducing circulating ALT levels, liver brightness and IR in NAFLD children. Despite the amelioration in serum markers of liver function and insulin sensitivity trended to be more evident in the vitamin treated arms, statistical significance was not achieved at the post hoc analysis, likely because of a large variability in the individual response and/or a small number of subjects in the not weight loser groups. Extreme variability is confirmed by large deviations mainly in ALT and HOMA-IR, as evident in per cent changes, which were not statistically significant, Table 1, and Figures 2 and 3.
We stratified our series by using a double approach, which basically produced similar results. Patients who lost either ≥20% of the EBW or ≥1.0 kg were separated from those children who did not. Since compliance and retention to the experimental programme was excellent, obese and over weight children were approximately differentiated from normal weight subjects (groups A2 and B2). Few over weight patients entered the weight non-loser groups (Figure 1) without significantly affecting results (data not shown). Improvements in ALT levels and HOMA-IR in obese children who lost weight in the placebo arm (group A1) are likely to reflect the effect of weight loss; while changes in group B1 should take into account the addictive effect of antioxidant therapy. In subjects who did not lose weight since thye were almost normal weight (groups A2 and B2), the improvement in liver function and brightness may be because of the effect of both physical exercise and healthy diet low in high-glycaemic index carbohydrates and saturated fats, as previously demonstrated.2 Antioxidant therapy did not cause any significant additive improvement in liver function and insulin sensitivity. In obese and overweight children a decrease of as little as 1 kg was sufficient to induce a significant reduction in ALT levels and amelioration of insulin sensitivity. In agreement with our findings, Hickman et al. found that the parallel decrease in ALT and insulin levels was associated with the amount of weight loss in chronic liver disease. A sustained improvement in ALT and insulin levels was seen with a weight loss of as little as 4–5% body weight, even without normalizing BMI.34
As far as the effect of vitamin E on liver function and IR concerns, to the best of our knowledge, this is the only study looking at the effect of antioxidant therapies on IR in children with liver biopsy-proven NAFLD, while a number of studies have been performed to evaluate the effect of the vitamin on liver function alone in paediatric and adult NAFLD.
In children, an open-label pilot trial of antioxidant therapy with doses of vitamin E ranging from 400 to 1200 IU daily was initiated to treat 11 obese patients with elevated aminotransferase levels without evidence of other liver disease.8 During a mean follow-up of 5.2 months, despite the decrease in body weight was insignificant (BMI from 32.8 to 32.5 kg/m2) as well as the compliance to the diet advices, the author observed the normalization of both aminotransferase levels during the treatment, but no improvement in liver brightness. Drug was titred according to the individual response. Patients were monitored after withdrawing vitamin E therapy, and they experienced an increase in liver enzyme concentrations. Vajro et al.9 found no apparent beneficial effect of vitamin E supplementation in children who effectively lost weight, while the authors recommend the antioxidant prescription in those patients who did not lose weight. The authors did not observe any changes in liver brightness, despite the normalization of liver enzymes occurred already after 2 months of therapy.
In adult NAFLD, vitamin E, 300 mg day, significantly reduced transforming growth factor-beta concentrations, a pro-fibrotic cytokine, in patients, who had already gone through a life-style modification programme, with an improvement of liver fibrosis in some patients.11 Harrison et al.10 found a significant, albeit modest improvement, which was independent of the amount of weight loss, in fibrosis within the vitamin group after 6 months. Nevertheless, the authors also observed a significant drop in ALT in the placebo group, despite a trivial decline in BMI. A couple of studies, mainly evaluating the efficacy of weight loss in patients with NAFLD and chronic hepatitis C, have shown no fibrosis improvement, but amelioration in steatosis and liver function tests.35, 36 A short-time therapy with 800 IU vitamin E did not modify levels of ALT and pro-inflammatory cytokines (IL-8 and tumour necrosis factor-alpha) in 16 biopsy-proven NAFLD patients.37 We found that the insulin sensitivity increased to an extent not different among vitamin and placebo treated arms.
Kugelmans et al.37 found that vitamin E improved insulin sensitivity and several of its associated parameters, including ALT levels in overweight otherwise healthy subjects, but the effect of treatment was not sustained over the time.37 In type 2 diabetic patients, circulating levels of ROS were reduced and glycemic control improved by the vitamin E administration.38–40
The strengths of our study are that it is first one conducted in the largest series of biopsy-proven paediatric NAFLD patients and the strong compliance and retention of patients to the experimental programme, but several caveats must be considered. First, the lack of any information on post-therapy liver histology must be pointed out, as there is no agreement on the relation between ALT levels and liver histology and evaluation of liver brightness is strongly dependent upon the operator. In two studies, serum aminotransferase levels correlate poorly10 or not reliably41 with histological disease activity. Thus, we cannot exclude an improvement in the liver inflammatory process. Second, we did not evaluate circulating levels of alpha-tocopherol at the baseline and during the treatment. The dose of antioxidants prescribed could be considered not ‘pharmacological’ as compared with the amount of alpha-tocopherol utilized in previous paediatric studies.8, 9 Third, there was no way to check the adherence to the physical exercise programme.
In conclusion, our results suggest once again that simple life-style intervention with diet and physical exercise in children with NAFLD can lead to a significant improvement of liver function, glucose metabolism, lipid levels beyond any antioxidant therapy. We strongly believe that changes in life-style, because of the tight follow-up who underwent all patients enrolled, might have significantly affected dietary habits, body composition through the increase in physical activity, also in those patients who did not lose weight. Therefore, changes in life-style should represent the first step in the management of children with NAFLD. It remains uncertain however, whether the addition of medications that improve antioxidant defences in the liver would be more efficacious than life-style intervention alone.