Despite substantial scientific interest in dietary influences on NAFLD, little information is available about diet in children with NAFLD compared to the histologic features of the disease. In this registry based study, we were able to examine self reported diet and compare this to rigorously measured histology. We found that diet (after the diagnosis of NAFLD) did not differentiate simple steatosis and NASH. We also found several interesting associations that suggest areas for further investigation, including very low consumption of vitamin E, weak associations between vitamin E and C and increased histologic severity and finally, strong associations between uric acid (a surrogate marker for fructose consumption) and histologic severity.
We did not find any significant differences in diet when we compared children with “NASH” to “not NASH.” Each group reported similar consumption of fat, sugar -sweetened beverages, antioxidants and other micronutrients. Diet comparisons by NAS score, representing a composite of these findings, also did not differ significantly. These findings suggest that diet is not the primary cause of whether a child with NAFLD has NASH or not.
In our cohort of children with NAFLD, the median reported consumption of vitamin E in all groups was less than half of the US recommended daily allowance of Vitamin E (for adolescents 22.5 IU/day)(11
). Because of the registry design, our study did not include obese children without NAFLD so we do not have a disease-free control group for comparisons. However, in general, obese children have not been identified as having lower reported intake of vitamin E and C. Data from the National Health and Nutrition Examination Survey (1988-1994) demonstrated that children who were obese had similar reported intake of these vitamins as well as fruits and vegetables compared to non-obese children.(12
In our study, we found weak associations with vitamin E and C consumption compared to steatosis and ballooning respectively. Both, vitamin C and vitamin E function as scavengers of hydoxyl, peroxyl and superoxide radicals and protect against plasma lipid and low-density lipoprotein peroxidation(13
) and other oxidative stress and thus could be important in preventing the progression of NAFLD. Vitamin E has been tested as a treatment for NAFLD(14
) in part because antioxidant deficiency may lead to increased lipid peroxidation and cell death due to mitochondrial compromise (evident as ballooning on liver biopsy). In the recently published PIVENS randomized, controlled -trial comparing pioglitazone, vitamin E and placebo in adults with NASH, vitamin E therapy was associated with improved ALT, AST, steatosis, lobular inflammation and ballooning(18
), suggesting a protective role in hepatocytes. In the TONIC treatment trial of children and adolescents with NAFLD, 58% of those with Vitamin E had histologic resolution of NASH, with significant improvement over the 28% with NASH receiving placebo (19
There are several previous non-pathology based published studies that also examine the diet of children with and without presumptive NAFLD based on surrogate markers. Quiros-Tejeira et al found a small increase in consumption of dietary cholesterol in the suspected NAFLD children compared to than normal ALT subjects(20
). De Piano et al studied 43 adolescents, including 13 with NAFLD (based on ultrasound evaluation) and found no significant differences in total energy, % protein, % carbohydrates, % fat or cholesterol consumption compared to obese adolescents without echogenic livers(6
). Panpodreou et al compared adolescents with and without NAFLD (using ultrasound) and total energy, % protein and % fat were similar. However, they found an increase in carbohydrates and sugar intake in children with NAFLD(21
We were also interested in examining sugar consumption in our subjects because added sugars are known to be associated with dyslipidemia(22
) and fructose can be used to induce fatty liver in animal models(23
). Adolescents are the highest consumers of both added sugars and fructose, making them a high risk group for potential effects(24
). Because the Block Brief questionnaire lacks a detailed breakdown of sugar, we utilized sugar sweetened beverages (the largest source(24
)) as a surrogate marker of fructose intake. Interestingly, we did not find any difference in reported sugar sweetened beverage consumption between groups. Previous studies of adult NAFLD patients, including a registry study using a similar design to ours(25
), have found increased reported intake of sugar sweetened beverages, elevated uric acid levels (26
) and associations between uric acid and fibrosis severity(25
). Uric acid levels increase with fructose intake(30
) and intake of fructose correlates with uric acid levels in the general population(31
). Because of this it has been used as a surrogate marker of fructose intake. Despite the lack of difference in reported sugar sweetened beverage consumption, our groups differed significantly in uric acid level with the highest levels found in the definite NASH group (p=.008). When definite NASH was compared to steatosis alone (not NASH), there was a trend towards a higher uric acid level in those with definite NASH (p=.07). Because sugar sweetened beverages only account for an average of ~40% fructose in the diet(24
), our subjects may have substantial fructose intake from other sources (such as processed foods) that we were unable to measure given the limitations of our dietary instrument. In addition, it is possible that uric acid has an independent effect in NAFLD, unrelated to fructose intake. Further studies with both histology and more detailed diet information will be needed to understand the relationship of fructose to NAFLD and more specifically to NASH in children.
A possible limitation of our study may be a reporting or recall bias because most participants reported a relatively healthy diet. For example, median sugar sweetened beverage intake was reported as 1 glass or can or less per week. In the United States, average intake of sugar sweetened beverages for children age 12 to 19 years represented 356 calories per day(32
), which would translate to 2.5 cans of a typical 12 oz can of soda or more than three 8 oz glasses of fruit-like drink per day. Thus, our subjects report a much lower than average consumption of sugar sweetened beverages. In addition, less than one-fourth of our subjects reported >40% fat intake, the definition of a high fat diet. There are several factors that may account for this finding. All of our participants had already undergone a substantial medical procedure (liver biopsy) and had been given a diagnosis of NAFLD. This event may have been an effective trigger for lifestyle improvements, including alterations in diet. At the time of diagnosis, many patients are instructed to decrease sugar sweetened beverages, increase fruits and vegetables and reduce fat intake as part of standard therapy. Most of our subjects were evaluated at 1-2 months after the liver biopsy, possibly a peak time for implementing an improved diet. The Block Brief Questionnaire asks participants to reflect on their diet over the past year, however the recall may be more influenced by their current diets or they may wish to appear in conformance with recently provided nutritional advice for a healthy diet. Overall, these effects may be less important because it would likely affect all participants, regardless of histologic variation.
Several of our findings had p values of .05. Use of multiple comparisons could lead to p values identified as significant that are actually random. However, there is a pathophysiological basis for these findings, and the results support work by Strauss et. al. that obese children are low in vitamin E (12
) and the PIVENS trial. In our study, the trends for uric acid, vitamin E and vitamin C were not consistent across the 3 groups. This is likely as result of the difficulties of assigning scores and cutoffs to continuously variable pathology findings. An alternative hypothesis is that vitamins may have threshold levels and are associated with worsened pathology only when they fall below certain levels.
The Block Brief questionnaire has inherent limitations. Cullen et al compared a similar Block survey (Block Kids) in children to two 24 hour dietary recalls and found that the Block overestimated the percent energy from carbohydrates and found significant differences in the means for most food groups and nutrients between the two methods(33
). They found that the Block was more accurate for children > 12 years of age and for nutrients (compared to food groups), both bolstering the findings in this study since our mean age was > 13.0 years and we examined specific nutrients, not food group servings. Mexican American diets are not well-represented in the Block Brief Questionnaire, and some of our subjects are Mexican Americans.
In summary, macronutrients did not differentiate between simple steatosis and NASH in the children in our study. We found that children with NAFLD consumed less than the recommended amounts of vitamin E and that there was a weak association between lower consumption of both vitamin E and C and pathologic severity of NAFLD. Uric acid, a surrogate marker of dietary fructose, was significantly increased in those children with definite NASH compared to milder forms of NAFLD. Prospective studies are needed to evaluate diet in potential subjects prior to diagnosis and nutritional counseling (and ideally before NAFLD onset) in order to better confirm dietary contributors to this disease.