Obesity is an increasing problem in the developed world, in children as well as adults. Nonalcoholic fatty liver disease (NAFLD), most commonly seen in the context of obesity and insulin resistance, is a common cause of liver disease in all age groups. It is estimated that 10% of obese children manifest fatty liver disease as reflected by increased alanine aminotransferase (ALT).1 NASH (nonalcoholic steatohepatitis) has been shown to cause cirrhosis in children as young as 12 years old.2 Treatment options are limited, although gradual weight loss is probably beneficial. In order to conduct therapeutic trials, however, the disease must be carefully defined so that disease remission or progression can be measured.

The study by Schwimmer et al. in this issue of HEPATOLOGY reports the largest pediatric NAFLD histopathology series to date.3 One hundred consecutive liver biopsies from children ages 2 to 18 years were reviewed. Ninety-eight percent of the children were either obese or overweight and 8% were diabetic. The pathological spectrum seen ranged from simple steatosis (only 16%) to NASH (84%). Eight percent had grade 3 or 4 fibrosis with cirrhosis in 3%. A mathematical technique called agglomerative cluster analysis was used to analyze histological patterns and characterize distinct phenotypes. The authors proposed two subtypes of liver injury in pediatric NASH. Type 1, accounting for only 17% of the children in the series, was characterized by steatosis with ballooning degeneration and/or perisinusoidal fibrosis, without portal features. This type 1 pattern, seen in most adults with NASH, was uncommon in this series of children. Fifty-one percent of the children displayed type 2 NASH, defined by steatosis with portal inflammation or fibrosis without the bal- looning degeneration and perisinusoidal fibrosis seen in classic NASH. Clinical and demographic characteristics clustered within the two groups: the children with type 2 (portal) NASH were younger, heavier, and more likely to be male and non-white.

The technique of agglomerative hierarchical cluster analysis is used to group objects in a way that maximizes similarities between objects within a group and emphasizes differences between groups.4 The hierarchical tree in Fig. 1 of Schwimmer et al. begins with the individual patients and then clusters them into broader and broader groups (with increasing dissimilarities) until the larger types 1 and 2 are identified. Differences between these groups can be analyzed, along with, as the authors state, “biological meaningfulness.”

Clearly these findings need to be substantiated in the biological context of larger and more heterogeneous groups of patients. Schwimmer's patients were 14% white non-Hispanic and 67% Hispanic. The population seen in this San Diego pediatric program is not representative of children with NAFLD seen in most other centers. Larger studies are needed to delineate this disease in the non-obese, the diabetic, and the hyperlipidemic child. Data from different ages and stages of the disease are needed to validate the clinical significance of the type 1 and 2 clusters.

Previous studies have recognized portal involvement in patients with NASH, both adult and pediatric. Abrams et al., for example, found “isolated portal fibrosis” in 33% of patients in his adult bariatric surgery series.5 Several authors have noted patterns of portal injury and fibrosis in pediatric NAFLD. In Baldridge's multicenter retrospective study of NASH in children, 13 of 14 patients had portal fibrosis.6 Rashid and Roberts illustrated portal fibrosis and inflammation in his report on 36 children with NASH,7 and Moran et al.8 and Kinugasa et al.9 also report portal fibrosis in their pediatric series. In their HEPATOLOGY report on the development of a histological scoring system for the NASH Clinical Research Network, Kleiner and Brunt and colleagues compared evaluations by nine study pathologists on 18 pediatric and 32 adult NAFLD liver biopsies.10 The study pathologists noted that pediatric cases had more periportal steatosis, less ballooning, less Mallory's hyaline, and more “periportal only” fibrosis. Isolated periportal fibrosis was four times more common in children than in adults in these biopsies. The relative absence of the “classic” NASH lesions of ballooning or lobular inflammation made it difficult to generate a definitive diagnosis of NASH in biopsies of children: only a third of the biopsies in this series could be characterized as “definite steatohepatitis.” Pediatric cases resulted in a lesser degree of interobserver agreement in scoring NAFLD features, making characterization difficult. The efforts by Schwimmer et al. to more clearly define pediatric NASH may assuage some of these problems.

This paper brings up several interesting issues of physiology. The percent steatosis, for example, is nearly the same in the two groups in spite of such different histology. As the authors speculate, something else (hormones, ethnic factors) is influencing these two histological phenotypes. Oddly, in some children, the fibrosis occurs in a (portal) site remote from most of the steatosis. The paper alludes to the possibility that steatosis and fibrosis are not directly causally related. Perhaps hepatic metabolic zonation in children may have a different pattern, resulting in a more common second form of fatty liver disease. We must reconcile, however, the fact that a number of adults have been reported with this same type 2 periportal pattern of NASH. This pattern is much more common in children but is hardly unique.

While the authors state that more severe fibrosis is seen in type 2 NAFLD, we have no longitudinal data regarding the clinical behavior and natural history of the two types. It is not certain that NASH neatly progresses from simple steatosis to steatohepatitis to fibrosis to cirrhosis.11 Type 1 and type 2 NASH may even be different stages of the same disease. This question of whether these are truly different clinical types of NASH is crucial. If, as the authors propose, the pathophysiology of the two forms is distinct, then they may have different natural histories. Effective treatment may differ as well. In a large multicenter study like the NIH-sponsored NASH Clinical Research Network (CRN), armed with these two descriptions, we can assess natural history and therapeutic interventions with the possibility in mind that we are dealing with two different diseases, not just two different histological patterns.

As every pediatrician will tell you, “kids are not just small adults.” The paper by Schwimmer et al. demonstrates this at the histological level and the Kleiner paper makes note of the fact that children are different enough that they often do not even receive a definitive diagnosis of NASH. Because liver biopsy is invasive, and because we lack standards of care for the role of the biopsy in evaluating and following serious liver diseases in children, we have limited data to use in the decision-making processes that are becoming more and more common as the obesity epidemic in children continues. We need to know more about children with NAFLD, and liver histology is vital because laboratory studies may not reflect liver injury. It is notable that several patients in the Schwimmer report had normal ALT with type 1 or 2 NASH, two of Rashid's patients had moderate to severe fibrosis with no inflammatory activity, and two of Baldridge's patients had portal inflammation and fibrosis but normal ALT.3, 6, 7 Liver biopsies of children with NASH can help us follow-up on Schwimmer's findings of 2 different NASH phenotypes by tracking the actual disease process in well-defined groups. Several questions must be considered to move forward. When is liver biopsy for diagnosis of suspected NASH in a child appropriate? It is difficult to make progress in evidence-based therapy without characterizing the disease. What is the role of biopsy for follow-up? Biopsy findings must be interpreted with some knowledge of the progression of disease. What about normal-ALT NAFLD? We may be missing serious liver disease in obese children. More data are needed to answer these questions. Our understandable reluctance to submit children with NAFLD to the risks of biopsy may render us unable to help them. There is a bright spot, however. In the past, children with liver disease have had to wait for years for adult studies to clarify the natural history (in adults, that is) or evaluate treatments (such as interferon for viral hepatitis). The NASH CRN introduces a new paradigm where children are included in the large research network from the start, and their differences are factored in on the ground floor. Schwimmer's report gives us a valuable tool with which to begin to study NAFLD in children.


  1. Top of page
  2. References