With a still increasing prevalence, NAFLD is, by now, thought to be the most prevalent liver disease worldwide [2
]. Overweight and insulin resistance are among the key risk factors for the development of NAFLD [34
]; however, despite intense research efforts, the question as to why some overweight individuals develop NAFLD and others do not is yet to be fully answered. In the present study, employing a cohort of randomly selected overweight children with no known signs of liver disease or other metabolic diseases before the study, it was shown that children with early signs of NAFLD, e.g., fatty liver grade 1 as assessed by ultrasound, had a higher BMI, BMI-SDS, and waist circumference than overweight children without NAFLD. Also, while still being within the normal range, AST activity was higher in the sera of overweight children with NAFLD, whereas neither ALT activity nor markers of glucose metabolism nor other metabolic markers, such as blood pressure and triglycerides, differed between groups. Somewhat contrasting the findings for waist circumference, the ratio of adiponectin to leptin, suggested to be indicative of visceral fat volume and NAFLD in obese adolescents [36
], was similar between overweight groups. However, results of a recently published study by Dhaliwal et al. [37
] using computed tomography to assess hepatic steatosis and the abdominal fat area suggest that, while abdominal subcutaneous adipose tissue per se is greater in children with hepatic steatosis than in those without, increases in visceral adipose tissue area seem to be related to the presence of steatosis in older children (≥9.8 years). Therefore, the apparent lack of relation of waist circumference and the ratio of adiponectin to leptin might be related to the rather young age of study participants (<9 years) and to differences in subcutaneous adipose tissue mass rather than visceral adipose tissue. Furthermore, results of others also suggested that overweight children with hepatic steatosis have a higher BMI and waist circumference when compared to overweight children without signs of NAFLD [5
]. Previous studies of others also reported that transaminase activities in the sera of overweight children were also similar to those of overweight children serving as controls [38
]; however, in contrast to the findings of the present study, in studies of others, a strong association of the presence of NAFLD in overweight children with increased markers of insulin resistance, dyslipidemia, and the presence of metabolic abnormalities was found [5
]. Indeed, in the present study, total cholesterol in serum was even found to be higher in overweight children without NAFLD than in those with NAFLD. Differences between the results of these studies (>9 years old) and the present study (<9 years old) might have resulted from differences in the age of study participants, as well as in the severity of steatosis (in the present study, only grade I vs. minimal-to-severe fatty liver or even beginning non-alcoholic steatohepatitis (NASH) in other studies) and obesity of participants enrolled [5
]. The study setting (here, children recruited in schools vs. hospitals in most other studies) and markers used to assess insulin resistance (in the present study, fasting glucose and insulin vs. oral glucose tolerance tests) were also markedly different between these studies and the present study [5
]. Reasons for the significantly higher levels of total cholesterol and LDL cholesterol in the sera of overweight children without NAFLD have to be delineated in future studies.
Also, while overweight children with NAFLD only showed very early signs of the disease, IL-6 and TNFα levels in plasma were both higher in children with NAFLD than in overweight controls, whereas CRP and active PAI-1 levels in plasma—both markedly higher than in normal-weight controls—did not differ between overweight groups. In line with these findings, others showed that concentrations of IL-6 and TNFα in serum are closely related to the prevalence of NALFD [40
]. Results of others also suggest that, in overweight children, CRP and active PAI-1 levels may be elevated independently of the presence of NAFLD [41
]. Indeed, while active PAI-1 levels were shown to be related to severe stages of the disease, e.g., manifest steatosis or NASH [43
], this acute-phase protein was also shown to be related to HOMA-IR by others [44
], and it was similarly higher in both overweight groups in the present study when compared to controls. Taken together, results of the present study suggest that, in overweight children, very early stages of NAFLD are associated with higher body weight, greater waist circumference, and elevated proinflammatory cytokine levels while, markers of insulin resistance are not different. However, the results of the present study by no means preclude that an impaired glucose tolerance or insulin resistance contributes to the onset of NAFLD. Indeed, in adults and mouse models, it was shown that both fasting insulin and glucose levels can still be within the normal range in peripheral blood, while, in liver tissue, the expressions of insulin receptor and insulin receptor substrate were markedly lower [45
]. Therefore, it could be that, in the present study, overweight children with NAFLD may have suffered from impairments of insulin signaling and glucose metabolism in liver tissue, while fasting glucose and insulin concentrations in peripheral blood were still within the normal range. This needs to be addressed in future studies.
4.1. Absolute Energy Intake, Nutritional Intake, and Dietary Pattern of Overweight Children with and without NAFLD Differ
Results of animal studies suggest that not only general overnutrition, but also the composition of the diet, e.g., the proportion of saturated fatty acids and sugars, and herein, especially of fructose, may be critical in the development of NAFLD [19
]. In a cohort of children with NAFLD, Mosca et al. [33
] recently showed that dietary fructose intake is independently associated with NASH. Furthermore, it was shown that children with NAFLD absorb and metabolize fructose more effectively than normal-weight children [48
]. In the present study, overweight children with early signs of NAFLD had a significantly higher mean daily total energy intake when compared to overweight children without NAFLD (~250 kcal/day) which mainly seemed to result from a higher daily total fructose (free fructose and fructose derived from sucrose) and total glucose (free glucose and glucose derived from sucrose) intake originating from a markedly higher soft-drink and juice intake. Results of the present study are in line with the findings of others, showing that both children and adults with NAFLD have a higher mean fructose intake mainly resulting from a higher consumption of soft drinks and fruit juices [10
]; however, in most of these studies, normal-weight healthy individuals were compared with overweight patients with NAFLD [10
]. Indeed, the number of human studies comparing the nutritional intake and dietary pattern of overweight individuals, and even more so, weight-matched individuals with and without NAFLD is rather limited. In line with the findings of the present study, Ouyang et al. and Assy et al. [52
] showed that adult patients with NAFLD drank more soft drinks and juices.
Interestingly, normal-weight children enrolled for comparison almost had similar daily total energy, monosaccharide, and disaccharide intakes without showing any signs of NAFLD or other metabolic diseases when compared to overweight children with NAFLD. Yet, normal weight children, on average, were ~16 h/week (~140 min/day) sedentarily active doing handcrafting, drawing, reading, and watching TV or playing video games, while overweight children with NAFLD, on average, spent 25 h/week (~215 min/day) with these activities. Indeed, results of Felix et al. [54
] suggest that, in overweight children, the intake of refined carbohydrates and the lack of physical activity were associated with a higher risk of developing NAFLD, further suggesting that protection against the development of NAFLD, and probably, overweight in normal-weight children was strongly dependent upon their physical activity in the present study. In support of this hypothesis, lifestyle changes not only focusing on changes in dietary habits, but also on increasing aerobic exercise were suggested to improve aminotransferase activity levels in children and adolescents with NAFLD [50
4.2. Overweight Children with NAFLD Have Higher Bacterial Endotoxin and LBP Levels in Peripheral Blood Than Overweight Children without NAFLD, Which Were Lowered by Moderate Dietary Counseling
In the present study, both bacterial endotoxin and LBP levels were significantly higher in overweight children with NAFLD than in those without. These findings are in line with the results of several human and animal studies, suggesting that alterations of intestinal barrier function, and subsequently, an increased translocation of bacterial endotoxin are critical in the development of NAFLD [10
]. While data derived from animal studies suggest that these alterations may be related to the intake of fructose [14
], results of human studies are somewhat contradictory. Indeed, in some studies, it was shown that both the intake of dietary fructose and bacterial endotoxin levels in peripheral blood were elevated in patients with NAFLD; however, frequently, not only fructose intake, but also total caloric intake of patients was significantly higher than that of controls [49
]. So far, studies focusing on a reduction of fructose intake suggest that, in adults with steatosis and steatohepatitis, this kind of counseling is associated with an improvement in liver status, a reduction in bacterial endotoxin levels, and improved intestinal barrier function [59
]. In the present feasibility study, while only slightly affecting overall weight and metabolic status, the moderate dietary counseling focusing only on a reduction in dietary fructose intake was associated with a reduction in bacterial endotoxin and TNFα levels, almost to the level of normal-weight controls, in overweight children with NAFLD. The apparent reduction in adiponectin levels from baseline to the end of study in children with NAFLD enrolled in the intervention arm might have resulted from the slightly, but not significantly, younger age of these children when compared to controls at baseline (7.5 vs. 8.0 years). Indeed, it was shown before that adiponectin plasma levels decrease between the ages of five and eight years [60
]. Furthermore, studies of Murphy et al. [60
] also showed that total cholesterol plasma levels increase over time in children aged between five and eight years, in line with findings of the present study. In contrast, despite lowering their BMI-SDS and maintaining their waist circumference, the bacterial endotoxin levels of overweight controls with NAFLD were unchanged. Taken together, these data suggest that the total intake of sugar-rich foods and bacterial endotoxin levels both may be critical in the development of NAFLD in overweight children. However, our results do not preclude that other factors such as genetic predisposition, intake of other nutrients, and sedentary lifestyle are also critical in the development of NAFLD. Rather, our data suggest that, at least in some children, targeting intestinal barrier function through dietary fructose intake may be beneficial in the prevention and therapy of this liver disease.
Our study is not without limitations which have to be considered when interpreting the results. Overweight children with and without NAFLD were not weight-matched; however, children were randomly recruited and enrolled in non-clinical settings, and, at the time of recruitment, had no known history of metabolic or liver diseases. Therefore, both overweight groups included metabolically healthy and unhealthy children. Thus, results might differ in larger and more homogeneous clinical studies. Nonetheless, as we aimed to study the early onset of the disease, this approach seemed to be the most feasible. Furthermore, the sample size of the intervention study was rather small, as the intervention focused only on children with NAFLD and five children were lost due to personal reasons or underreporting. Furthermore, as the focus of the intervention study was to show feasibility of this kind of moderate dietary intervention, no power calculation was performed to determine the number of subjects needed to be included for statistically significant outcomes. Thus, the characteristics of the feasibility study are rather explorative, and the effect of a moderate dietary intervention on metabolic and inflammatory markers needs to be assured in a larger randomized population. However, despite the small sample size, our findings are in line with others showing that dietary counseling might be beneficial in improving metabolic parameters in children [61
]. Furthermore, a selection bias cannot be ruled out as control and intervention groups were self-selected due to incompliance of many guardians for randomization. Indeed, due to drop out and underreporting, the number of Asian participants deciding to undergo nutritional counseling was lower than the number of Asian children in the control group. However, as we were highly dependent upon the willingness of parents and children to attend the regular counseling meetings, from our perspective, this was the most feasible way of avoiding an even higher drop-out rate. Indeed, when enrolling children into the study, guardians repeatedly pointed out their unwillingness to participate in regular meetings, suggesting that it felt too straining. Accordingly, it was also not possible to obtain valid data regarding nutritional intake and dietary pattern at the end of the intervention. Therefore, it is not clear if the beneficial effects on bacterial endotoxin levels found at the end of the study resulted from a change in fructose intake or dietary pattern, or other factors. The role and impact of fructose on the beneficial effects found in the intervention group will have to be addressed in future studies. Reasons for this incompliance to participate in meetings might have been that our study was not situated in a clinical setting, and that children were thought to be healthy with the exception of being overweight/ obese before the study. Another limitation is the reduction of BMI-SDS in both groups of the intervention study. This, in part, might have resulted from the fact that guardians became aware of the potential health issues of their children. Indeed, some of the families might have changed additional dietary habits, and prolonged time spent physically active, while time spent sedentary active was reduced, without bringing this to our attention. Furthermore, in the region of Germany where our study was situated, “healthy” nutrition is part of the curriculum in elementary school and sometimes even in kindergarten. Therefore, it cannot be ruled out that at least some of the children in the intervention study and probably also their parents, e.g., during parent–teacher conferences and school enrollment, might have received additional training in regards to avoiding sugar-rich foods and to following a healthy lifestyle. Furthermore, in the present study, physical and sedentary activities were only acquired by questionnaires rather than activity monitors. Additionally, no follow-up was carried out to assess sustainability of the intervention on weight status, metabolic disorders, and associated proinflammatory alterations. However, we thought that the length of the study would be sufficient to test the principal hypothesis that, in children, a diet focusing only on a reduction in fructose intake may be a sufficient measure for reducing bacterial endotoxin levels and concentrations of proinflammatory cytokines. Long-term effects will have to be determined in larger randomized studies with a longer duration and follow-up.