Article Text
Abstract
Objective: Non-alcoholic fatty liver disease (NAFLD) is the most common cause of liver disease in obese youth. Lifestyle intervention was demonstrated to improve NAFLD but follow-up studies after end of intervention are lacking. Furthermore the necessary degree of overweight reduction for improvement of NAFLD remains unknown.
Methods: We examined standard deviation score of body mass index (SDS-BMI) and transaminases in 152 obese children with NAFLD diagnosed by ultrasound at baseline, at the end of a 1-year intervention and 2 years after baseline. Within-subject changes of these parameters were compared by participation in the intervention based on physical activity, nutrition education and behaviour therapy.
Results: In contrast to 43 children without lifestyle intervention, participation in lifestyle intervention (n = 109) was associated with a significant decrease of transaminases and overweight 1 and 2 years after baseline (1 year: alanine transaminase (ALT) −10 U/l (−14 to −6); aspartate transaminase (AST) −5 U/l (−7 to −3); SDS-BMI −0.23 (−0.30 to −0.16); 2 years: ALT −9 U/l (−12 to −6); AST −6 U/l (−7 to −4); SDS-BMI −0.30 (−0.37 to −0.22); data as mean changes and 95% confidence interval compared to baseline). Any degree of overweight reduction was associated with a significant decrease of NAFLD prevalence. The greatest decrease of NAFLD prevalence (1 year: −89% (95% CI −72% to −100%); 2 years: −94% (95% CI −83% to −100%)) was observed in children with the greatest overweight reduction (SDS-BMI decrease >0.5).
Conclusions: Multidisciplinary lifestyle intervention is effective to improve NAFLD even in the 1-year follow-up after the end of intervention. A minimal reduction of overweight led to an improvement of NAFLD.
Trial registration number: NCT00435734.
Statistics from Altmetric.com
Concurrent with the rise of childhood obesity,1 non-alcoholic fatty liver disease (NAFLD) is recognised as the most common cause of liver disease in obese youth.2–4 NAFLD in children was first reported in the early 1980s.5 Since then, a number of case series have been described with the following clinical characteristics: male predominance, higher relative increase in serum alanine transaminase (ALT) than in serum aspartate transaminase (AST), and a strong association with obesity.6–10 Although population prevalence is very difficult to establish, the prevalence of NAFLD in children was reported to be 2.6% in normal-weight children, while in previous reports on obese children and adolescents, it varied widely from 20% up to 77%.10–17 The complete metabolic phenotype and the pathogenesis of NAFLD remains to be established.2 A relationship between NAFLD and insulin resistance is postulated but discussed controversially.18 The commonly favoured two-hit hypothesis is comprised of an accumulation of fat (“first hit”) within the liver, which predisposes to the “second hit”, namely hepatocyte injury, inflammation, and fibrosis, the so-called non-alcoholic steatohepatitis.2 While the prognosis of NAFLD seems to be benign in the majority of children, development of chronic liver dysfunction and liver cirrhosis has been reported.5 19 20 Furthermore, development of hepatocellular carcinoma has been described in obese adults with NAFLD while this complication has not yet been reported in obese children with NAFLD.21
Studies in obese adults have shown that NAFLD may resolve with weight reduction.22 Similar findings following weight reduction have been reported in obese children and adolescents.11 23–25 However, patients with rapid and drastic weight loss were found to have developed slightly more portal inflammation and fibrosis.26
Despite the well-established short-term benefit of lifestyle intervention on NAFLD, follow-up studies after intervention are still missing questioning the long-term effect. Furthermore it is unknown which degree of overweight reduction is necessary to improve NAFLD. Therefore, we performed a longitudinal study in a large cohort of obese children with NAFLD. The primary aim was to determine the long-term effect of lifestyle intervention in obese children suffering from NAFLD. Additionally, we analysed exploratively the longitudinal relationships between overweight reduction and improvement of NAFLD as well as between the longitudinal changes of transaminases and insulin resistance in children with NAFLD.
METHODS
All obese children presenting at the outpatient obesity clinic of the Vestische Children’s Hospital, University Witten/Herdecke were screened for NAFLD according to the German guidelines for obese children.27 Obesity was defined according to the International Obesity Task Force using population-specific data.28 29 NAFLD was defined by presence of fat in the liver as “bright liver” in the abdominal ultrasound. Differential diagnoses were excluded in all children with suspected NAFLD by measuring serum creatine kinase, antinuclear antibodies (ANA), liver auto-antibodies (SMA, LKM, SLA), copper, caeruloplasmin, 24 h urinary copper, α1-antitrypsin, and EBV, HAV, HBV, HCV serology according to international recommendations.2 13
A total of 187 obese children with NAFLD were identified in the years 1999–2005. These children were encouraged to participate in the following study. Inclusion criteria were age 6 to 16 years and receiving regular school education. Exclusion criteria were endocrine disorders, premature adrenarche, syndromal obesity, any regular medication, and families with parents or children declaring no motivation or couldn’t find the time to attend regularly in the lifestyle intervention. All obese children were encouraged to participate in the 1-year outpatient lifestyle intervention programme “Obeldicks”, which has been described in detail elsewhere.30 31 Briefly, this outpatient intervention programme for obese children was based on physical activity, nutrition education, and behaviour therapy including individual psychological care of the child and his or her family (fig 1). The nutritional course was based on a fat- and sugar-reduced diet as compared with the every-day nutrition of German children. The diet contained 30% fat, 15% proteins, and 55% carbohydrates including 5% sugar. As control group for the lifestyle intervention served all obese children who were not able to participate in the “Obeldicks” training programme because they lived too far away and had no means of transport. These children and their parents were advised at their first presentation in a 15-minute consultation as to a suitable diet, the necessary physical exercise and behaviour patterns. They were given written information on nutrition and accompanying recipes. All children were requested to present at the outpatient obesity clinic 1 and 2 years after baseline for follow-up. Short advice concerning healthy lifestyle was provided during these examinations. The complete study recruitment procedure resulted in 160 obese children with NAFLD who agreed to participate (fig 2).
We examined weight, height, AST, ALT, insulin, and glucose concentrations in all children. These parameters as well as abdominal ultrasounds were determined at baseline, 1 and 2 years later. Ultrasound was performed by one investigator, who was blinded for the change of weight status. NAFLD was defined at baseline as well as in follow-up by abdominal ultrasound.
Height was measured to the nearest centimetre using a rigid stadiometer. Weight was measured unclothed to the nearest 0.1 kg using a calibrated balance scale. Standard deviation score of body mass index (SDS-BMI) was calculated according to German percentiles.29 We used the lambda-mu-sigma method to calculate SDS-BMI as a measure for the degree of overweight due to the skewness of body mass index (BMI) distribution.32
Blood sampling was performed in the fasting status at 8 am. Serum AST and ALT concentrations were measured using commercially available test kits (ALTL-, ASTPL- Cobas Integra 400 Roche Diagnostics, Mannheim, Germany). Insulin concentrations were measured by microparticle-enhanced immunometric assay (MEIA™, Abbott, Wiesbaden, Germany). Glucose levels were determined by colorimetric test using a Vitros™ analyser (Ortho Clinical Diagnostics, Neckargmuend, Germany). Intra- and interassay correlation coefficients (CVs) were <5% for insulin and glucose measurements. Additionally, homeostasis model assessment (HOMA) was used to detect the degree of insulin resistance.33 The resistance can be assessed from the fasting glucose and insulin concentrations by the formula: resistance (HOMA) = (insulin (mU/l) × glucose (mmol/l))/22.5.
Statistical analysis was performed using the Winstat® software package and the statistical software package SAS version 9.1.3 (SAS Institute Inc., Cary, North Carolina, USA). Analysis of variance (ANOVA) for repeated measurements was used to examine differences between time points for obese children with and without lifestyle intervention. If the sphericity assumption was violated (Mauchly’s test of sphericity: p<0.05), Huynh–Feldt correction was used for estimating p values. The overall within-subject effect of the intervention was estimated in a doubly-multivariate repeated-measures ANOVA (MANOVA for repeated measurements). Post hoc tests compared means of different time points and were adjusted for multiple testing.
Additionally, children with lifestyle intervention were divided into four groups according to their degree of weight loss (increase of SDS-BMI, decrease of SDS-BMI 0–<0.25, decrease of SDS-BMI 0.25–0.5, decrease of SDS-BMI >0.5) and similarly analysed. We used this separation since in previous studies, an improvement of cardiovascular risk factors and insulin resistance was only detectable if SDS-BMI decreased >0.5.34 35 A reduction of 0.5 SDS-BMI is approximately a reduction of 2 kg/m2 BMI in growing children or a reduction of approximately 5 kg in adolescents who have reached their final height.
Pearson correlations were calculated between baseline liver enzymes, insulin, insulin resistance index HOMA, and changes of weight status. Furthermore, Pearson correlations were calculated between changes of liver enzymes, changes of weight status (SDS-BMI), changes of insulin, and changes of insulin resistance index HOMA. Significance level was set to p values <0.05. Data were presented as mean and 95% CI.
Written informed consent was obtained from all children and their parents. The study was approved by the local ethics committee of the University of Witten-Herdecke.
RESULTS
A total of 109 obese children with NAFLD participated in the lifestyle intervention and 51 obese children with NAFLD were divided into the control group. The drop-out rate was 7% in the intervention group and 16% in the control group (fig 2). The eight drop-outs of the control group were not available for follow-up. They did not differ in respect of age, gender, degree of overweight, or transaminases from the other children in the control group. All eight children dropping out of the intervention group did not reduce their overweight. Complete follow-up measurements were available in the eight drop-outs of the intervention group.
Eighty-two (75%) of the 109 children with lifestyle intervention reduced their SDS-BMI. The mean change of their SDS-BMI at 1-year follow-up ( = end of intervention) was −0.23 (95% CI −0.30 to −0.16) and the mean change between baseline and 2-years follow-up was −0.30 (95% CI −0.37 to −0.22). Participation in the lifestyle intervention was associated with a significant overweight reduction at the end of intervention and 2 years after baseline in contrast to children without lifestyle intervention (table 1). Obese children participating in the lifestyle intervention did not differ significantly at baseline with respect to age (mean age 11.1: 95% CI 6.2 to 15.7 years vs 11.6: 95% CI 7.9 to 15.8 years), gender, degree of overweight (SDS-BMI), AST, or ALT compared with children without lifestyle intervention (table 1).
Transaminases decreased significantly over time in children with lifestyle intervention but not in children without the intervention (table 1). In children with lifestyle intervention, ALT decreased significantly with −10 U/l (95% CI −14 to −6), AST with −5 U/l (−7 to −3) and NAFLD with −50% (95% CI −40 to −60%) between baseline and 1-year follow-up. The decreases between baseline and 2-year follow-up were −9 U/l (−12 to −6) in ALT, −6 U/l (−7 to −4) in AST and −45% (95% CI −41% to −49%) in NAFLD prevalence. When comparing the transaminases ALT and AST by intervention status, a significant decrease was observed in children participating in the lifestyle intervention (stratified MANOVA within subjects p<0.001), while no decrease could be observed in children not participating in the intervention (stratified MANOVA within subjects p = 0.650). Furthermore, the prevalence of NAFLD decreased significantly in children with lifestyle intervention (table 1). The improvements of transaminases were maintained even 1 year after the end of the intervention. However, post hoc tests among children participating in the intervention showed that this decrease was mainly due to the decrease in the first year (stratified MANOVA p<0.001 adjusted for multiple testing), while no change could be observed for the difference between 1 year and 2 years after the intervention (stratified MANOVA p = 0.307 adjusted for multiple testing).
Transaminases tended to normalise in lifestyle intervention even in the children with very low overweight reduction (SDS-BMI reduction >0–0.25) (table 2). SDS-BMI reduction of >0.5, >0.25–0.5, and >0–0.25 was associated with a significant decrease of transaminases (stratified MANOVA within-subjects effect p value of <0.001, 0.002 and <0.001). Post hoc tests among children with SDS-BMI reduction showed that the decrease of transaminases was mainly due to the decrease in the first year (stratified MANOVA p<0.001 adjusted for multiple testing), while a change could not be observed in the difference between 1 year and 2 years after the intervention (stratified MANOVA p = 0.716 adjusted for multiple testing). The largest improvement of transaminases and the largest decrease of NAFLD prevalence (−89% (95% CI −72% to −100%) at the end of intervention and −94% (95% CI −83% to −100%) in the 1-year follow-up after the end of intervention; table 2) was observed in the children with the highest degree of overweight reduction (SDS-BMI reduction >0.5). In contrast, an increase of SDS-BMI in lifestyle intervention was not associated with a significant change of transaminases (stratified MANOVA within-subjects effect p value = 0.328) or NAFLD prevalence (table 2).
In the summarised collective of 152 children with and without lifestyle intervention, baseline transaminases were not significantly correlated to baseline insulin or insulin resistance index HOMA (table 3). Additionally, baseline AST and ALT were not correlated to changes of SDS-BMI at 1-year and 2-year follow-up.
The changes of SDS-BMI in the course of the first year were significantly correlated to changes of ALT (r = 0.24), but not to changes of AST (r = 0.09). Similar findings were observed between changes of SDS-BMI, AST and ALT in the 2-year follow-up period (data not shown). Furthermore, changes of transaminases after 1-year and after 2-year follow-up were not significantly correlated to changes of insulin or insulin resistance index HOMA (table 3).
DISCUSSION
This is the first large longitudinal study in obese children with NAFLD demonstrating the effectiveness of a 1-year lifestyle intervention programme not only at the end of intervention but also 1 year after the end of lifestyle intervention.
The highest degree of overweight reduction was associated with the highest decrease of transaminases and prevalence of NAFLD. Furthermore, changes of ALT were correlated to changes of SDS-BMI demonstrating the impact of overweight reduction on levels of transaminases. Additionally, we observed a normalisation of liver enzymes in children with minimal overweight reduction. This finding is in concordance with the study of Tazawa.11
The observed changes in the transaminases and the prevalence of NAFLD in our sample represented the effects of a reduced caloric and fat intake as well as increased physical activity, which have been demonstrated in an earlier study by the participants of the “Obeldicks” intervention programme.36 Since physical activity, behaviour therapy, and nutritional education were performed together in the intervention group we cannot distinguish the impact of each of them on the prevalence of NAFLD. Furthermore, the effects of dieting and increased physical activity probably strengthened each other.
Lifestyle intervention was not associated with derangement of transaminases in our study. Even in the group of children with the highest degree of weight loss, no increase of liver enzymes was observed in contrast to studies in adults.26 37 However, in these reports the weight loss was achieved rapidly by gastric surgery, which is in contrast to the continuous weight loss over months in our study.
Although we used standardised procedures we were not able to achieve overweight reduction, normalisation of transaminases or decrease of NAFLD prevalence with recommendation of a suitable diet, the necessary physical activity and behaviour patterns but without the lifestyle intervention in the children with NAFLD. Long-term lifestyle intervention seems to be the therapy of choice in NAFLD due to the limited effectiveness of counselling only.
We did not find a significant correlation between transaminases and insulin or insulin resistance index HOMA both in cross-sectional as well as longitudinal analyses. Therefore our findings do not support the hypothesis of a close relationship between NAFLD and insulin resistance. However, we have to keep in mind that the HOMA model is only an assessment of insulin resistance and clamp studies are the gold standard to analyse insulin resistance. Furthermore, the levels of transaminases are only an indirect and limited parameter of liver injury. Additionally, based on the second hit hypothesis, insulin resistance plays a key role in a later stage of liver diseases: insulin is involved in switching from NAFLD to non-alcoholic steatohepatitis,38 which is unlikely to have occurred in most of our children.
The strengths of this study are its longitudinal design, the large study sample, and the control group of children with NAFLD without lifestyle intervention. However, this study has some potentially important additional limitations. First, this study was not randomised. We cannot exclude differing motivation or other influencing factors on outcome. It may be argued that carrying out a randomised study might be difficult for ethical reasons since such a study would necessarily prevent motivated obese children from being effectively treated. However, the children in the intervention group and in the control group did not differ significantly in respect to anthropometrical markers and transaminases. Furthermore, a significant impact of motivation on NAFLD seems unlikely. Second, the drop-outs in the control group may have influenced our results. However, at baseline, the drop-out did not differ in respect of age, gender, degree of overweight, or SDS-BMI from the study patients. Third, it cannot be ruled out that the children in the control group did not receive any further advice concerning their lifestyle. However, they stated at follow-up examinations that they did not participate in lifestyle intervention programmes. Fourth, BMI percentiles were used to classify degree of overweight. Although BMI is a good measure for overweight, one needs to be aware of its limitations as an indirect measure of fat mass. Fifth, this study included only hospital outpatients and not other community subjects. However, the obese children presented primarily to the obesity clinic because of their overweight irrespective of related comorbidities. Finally, diagnosis of NAFLD was not confirmed by liver biopsy. Liver biopsies in a large study collective such as our cohort are difficult to perform, also for ethical reasons, as no specific therapy follows histological diagnosis of NAFLD apart from recommending reduction of overweight which is generally advised to all obese children.
In summary, in obese children suffering from NAFLD, long-term multidisciplinary lifestyle intervention led to weight loss and an improvement of transaminases and prevalence of NAFLD in contrast to children without lifestyle intervention. Any degree of overweight reduction improved transaminases and the prevalence of NAFLD. Most importantly, the changes of weight status and the changes of transaminases and prevalence of NAFLD were sustained 1 year after the end of lifestyle intervention.
What this study adds
Multidisciplinary lifestyle intervention is effective to improve non-alcoholic fatty liver disease (NAFLD) not only at the end of intervention but also in a 1-year follow-up after intervention. Already a minimal reduction of overweight led to an improvement of NAFLD.
REFERENCES
Footnotes
Funding: Grant support from of the German “Competence Net Obesity”, which is supported by the German Federal Ministry of Education and Research (grant number 01 GI0839).
Competing interests: None.
Ethics approval: The study was approved by the local ethics committee of the University of Witten-Herdecke.
Patient consent: Obtained.