Introduction

Non-alcoholic fatty liver disease (NAFLD) is now the most common cause of chronic liver disease in both adults and children worldwide [1]. Increasing industrialization and commercial globalization in Asia, South America, and the Middle East have led to shifts toward more western dietary habits and reduced energy expenditure, which, in turn, have increased the prevalence of overweight and obesity, and consequently NAFLD as a common cause of liver disease, even in developing economies [2].

While diverse conditions can lead to abnormal hepatic steatosis, defined as steatosis in >5% of hepatocytes, NAFLD is predominantly associated with excess adiposity, in particular central adiposity, and occurs in the absence of significant alcohol intake. It can be found in lean individuals but typically in the setting of increased visceral adiposity or severe insulin resistance syndromes, such as lipodystrophy [3].

The term NAFLD encompasses a spectrum of fatty liver disease ranging from bland steatosis to steatohepatitis, referred to as non-alcoholic steatohepatitis (NASH). NASH is a potentially progressive and clinically more serious form of NAFLD that includes abnormal steatosis, histological evidence of hepatocyte injury (ballooning degeneration) and inflammation, with or without fibrosis. In adults, the pattern of injury and inflammation is typically localized to zone 3 of the hepatic lobule (termed definite NASH), though some portal inflammation can be present. Having some but not all of the features of definite NASH in a zone 3 pattern is classified as borderline zone 3 NASH. In children, portal predominant injury (zone 1 steatosis, portal and periportal inflammation, and periportal fibrosis) is more common, particularly before puberty. This entity has been termed borderline zone 1 NASH. NASH is more likely to be associated with development of fibrosis, primarily perisinusoidal in adults and primarily periportal in children in the early stages, but fibrosis can also be present in the absence of NASH.

In large cohort studies of patients diagnosed with NAFLD, both adults and children, 60–75% usually have bland steatosis with minimal inflammation and/or isolated mild fibrosis (termed non-alcoholic fatty liver or NAFL) but without the more prominent inflammation and ballooning injury characteristic seen in definite NASH. Depending on the pediatric cohort examined, definite NASH appears to affect up to 25% of patients with NAFLD and one-third may have borderline zone 1 NASH and of these 10–20% may have advanced fibrosis [4, 5]. Among children enrolled in a multicenter cohort study in the USA, 15% had advanced fibrosis (predominantly stage 3) at time of initial biopsy. In a population-based pediatric autopsy study, 9% had advanced fibrosis. The rate of progression of disease in children is not known. Case reports and series of children with rapid progression to cirrhosis have been reported, but these cases are not common. In contrast, among adults, up to 25% of patients can progress to cirrhosis within ten years of diagnosis.

Over the past decade, the public health impact of NAFLD has grown. NASH has emerged as the second leading cause of liver transplantation in adults in the USA and is anticipated to become the leading cause within the next decade [6]. Further, as rates of obesity, in particular severe obesity, increase worldwide, the prevalence of NAFLD is increasing globally [1]. This is likely to herald a significant international increase in rates of end-stage liver disease and liver-related mortality in coming decades and place ever greater stress on an inadequate supply of livers for transplantation [7, 8]. In the absence of easily implemented, widely effective pharmacotherapy, pediatric NASH is also likely to continue to contribute to this rise in end-stage liver disease and increasing need for liver transplantation in adults in the future. In fact, NASH is already the fastest growing etiology for liver transplantation in young adults, many of whom likely developed disease during childhood [9]. Although the consequences of hepatic steatosis, particularly when clinically silent in many individuals, may initially seem trivial in relationship to other pediatric liver diseases with more overt and acute complications during infancy and childhood, pediatric NAFLD rightly constitutes a significant public health burden whose prevention and treatment in childhood urgently need to be addressed.

Epidemiology: A Global Problem

Determining an accurate population-based prevalence of NAFLD remains elusive, as only liver biopsy can accurately distinguish between NAFL and NASH but is understandably implausible in epidemiological studies. The most robust population estimate of pediatric NAFLD prevalence in the USA remains a 2006 analysis of specimens derived from autopsies of 742 children who suffered accidental death between 1993 and 2003 in San Diego County [10]. The overall prevalence of fatty liver, defined as 5% of hepatocytes containing macrovesicular fat, was 13%, with the highest prevalence (17.3%) in adolescents aged 15–19 years. A disproportionately higher proportion of Hispanic (11.8%), Asian (10.2%), and white non-Hispanic children (8.6%) were affected compared with black non-Hispanic children (1.5%). In comparison, an estimated 31% of 2,287 adult participants from the Dallas Heart Study were found to have abnormal hepatic steatosis by MR spectroscopy [11]. Similarly, a disproportionately higher proportion of Hispanic (45%) and white adults (33%) were affected than black adults (24%), although the prevalence of abnormal steatosis in black participants was much larger than that found in the pediatric study (1.5%). Within the pediatric autopsy-based cohort with fatty liver, 23% had NASH, yielding an overall prevalence of NASH of 3%, similar to estimates among adult cohorts [10]. The relatively small and similar percentage of adults and children with fatty liver who develop steatohepatitis underscores the underlying genetic susceptibility to developing liver injury in response to hepatic steatosis in some individuals, whereas others are able to tolerate even large deposits of fat in the liver without developing a significant inflammatory or fibrotic response [12].

As the regional San Diego adolescent population was disproportionately Hispanic compared with the general USA, the authors subsequently adjusted the autopsy-derived estimate of 13% for age, gender, race, and ethnicity, yielding an overall estimated 10% prevalence of pediatric fatty liver disease in the USA. While this is the only pediatric population estimate based on histology, there are also several other pediatric population-based studies using serum alanine aminotransferase (ALT) levels and/or sonography as a screening measure. A 2015 systematic review and meta-analysis estimated a mean prevalence of NAFLD in the general pediatric population of 7.6% (95% confidence interval (CI) 5.5–10.3%, n = 20 studies) [13], but with considerable heterogeneity among studies. Similarly, the overall mean prevalence in the clinical obese population was higher at 34.2% (95% CI 27.8–41.2%, n = 56 studies), again with considerable heterogeneity across studies. Most studies used either ultrasonography or ALT as surrogate markers for NAFLD. Only two included histology data.

Serum ALT remains the most commonly used method to screen for NAFLD in clinical practice because it is inexpensive, minimally invasive, and widely available, and because there are currently no other validated serological biomarkers or panels of biomarkers that can accurately detect NAFLD in children. However, ALT has significant limitations in the detection of NAFLD. Firstly, serum ALT can be normal even in the presence of significant NAFLD. Nearly a quarter of children with histological NAFLD may have a normal serum ALT, and up to 60% of these may even have fibrosis [14]. Elevations of serum ALT are also not specific for NAFLD but can indicate a range of other possible causes of hepatitis, which need to be excluded before presuming a diagnosis of NAFLD. The definition of abnormal ALT also varies widely across different laboratories and populations. Local reference ranges are often calculated using regional populations that were not screened for silent liver disease, such as viral hepatitis, or obesity/overweight. As a consequence, many laboratory ranges for abnormal ALT are high and may miss clinically significant liver disease. In contrast, biologically derived norms for ALT in healthy lean individuals are typically much lower. In adults, for example, an upper limit of normal for ALT (defined as 95th percentile) derived from lean healthy Italian males and females was 30 U/L and 19 U/L, respectively. Similarly, the 95th percentile upper limit of normal for ALT in 982 metabolically normal, healthy weight adolescents aged 12–19 years in a National Health and Nutrition Examination Survey (NHANES) was 25.8 U/L for boys and 22.1 U/L for girls, far below the upper limits of normal typically derived in regional clinical laboratories (range 30–90 U/L in a representative survey) [15]. Similarly low upper limits of normal for ALT have also been found in a large multi-ethnic pediatric cohort in Canada [16]. In contrast, in most previously published studies using ALT as a screen for NAFLD, the defined abnormal level has ranged from >30 U/L to >40 U/L, which may have led to under-reporting of true population-based prevalence of fatty liver. However, an important caveat of using lower biologically derived thresholds for abnormal ALT in epidemiological studies is that it will improve the sensitivity for detection of NAFLD but lower the specificity for NASH. For example, applying the biologically derived cut-offs for pediatric ALT values to cohorts of children with documented hepatitis B or C infection and NAFLD doubled sensitivity for NAFLD to 85–92%, while decreasing specificity from 92% to 80–85% [15].

Across all screening modalities, the prevalence of NAFLD is strongly associated with increasing obesity, with the San Diego County autopsy study demonstrating a seven-fold rise in NAFLD prevalence as obesity increased, from 5% in normal, to 16% in overweight, and to 38% in obese children [10]. Varying rates of abnormally elevated serum ALT have been detected in cohorts of obese children. Higher estimates, approximately 40–50% prevalence, were found using ultrasonographic evidence of fatty liver as the screening method. In a 2019 study conducted in San Diego County and using MRI-determined hepatic fat content, a more precise method than ALT or ultrasound (US), the estimated prevalence of NAFLD in a cohort of 408 children with obesity was 26% (95% CI 24.2–27.7) [17]. Limited data exist on the prevalence of NAFLD in children with severe obesity. In a cohort of 148 adolescents who underwent liver biopsy at time of bariatric surgery, 59% had histologically confirmed NAFLD, with 34% of these having borderline or definite NASH [18]. However, this cohort was subject to referral and selection bias.

Additional risk factors for NAFLD identified in multiple population-based as well as smaller cohort studies include: insulin resistance, prediabetes and type 2 diabetes, Hispanic/Mexican-American ethnicity, male gender, and waist circumference. In one cohort of 134 multi-ethnic obese children, Hispanic children tended to present with NAFLD at younger ages, with 17.5% of Hispanic boys first diagnosed at younger than seven years of age. In addition, older age, elevated serum C-reactive protein and triglyceride levels were also positively associated with elevated serum ALT in a pediatric NHANES analysis. Many of these same clinical and demographic risk factors are also associated with increased histological severity or risk of fibrotic NASH in children, including older age, elevated triglyceride levels, Hispanic or Asian ethnicity, male gender, waist circumference, markers of insulin resistance, prediabetes and diabetes, and elevated blood pressure [19]. Elevated fasting triglyceride and glucose levels, waist circumference, hypertension, and obesity are considered components of the metabolic syndrome. Of these, increased waist circumference was the only component independently associated with liver fibrosis in a cohort of 197 Caucasian children aged 3 to 19 years with biopsy-confirmed NAFLD in Italy [20]. However, obese children with the metabolic syndrome were five times as likely to have any spectrum of NAFLD compared with obese children lacking the metabolic syndrome. The association of features of the metabolic syndrome with histological severity of NAFLD in children, in particular fibrosis, is aligned with data in adults that also suggest that metabolic syndrome increases odds of having severe fibrosis by 3.5-fold.

Diabetes is also emerging as a strong risk factor for more severe NAFLD. In a multicenter cohort of 675 children with biopsy-confirmed NAFLD, the odds of having NASH were significantly increased in those with prediabetes (odds ratio 1.9, 95% CI 1.2–2.9) and diabetes (odds ratio 3.1, 95% CI 1.5–6.2) compared to children with normal glucose [21]. Notably, both prevalence and severity of NAFLD appears to be decreased in black obese children, despite a high prevalence of obesity and heightened risk for cardiovascular disease risk factors including diabetes, suggesting potential for differing genetic risk factors.

The natural history of NAFLD, including risk of liver disease progression, has been difficult to determine since estimation of fibrosis requires a liver biopsy, and large-scale prospective natural history studies with serial biopsy data in children and adults have not yet been published. However, a pooled analysis of ten small case series in adults with longitudinal follow-up biopsies has estimated that up to 37% of adults with NASH may develop progressive fibrosis over a wide range of follow-up (1–21 years), with a smaller percentage (15–20%) developing severe fibrosis within a decade. Once cirrhosis develops, the risk of mortality does not appear to differ when compared with hepatitis C virus-associated cirrhosis [22]. Hepatocellular carcinoma can evolve from fibrotic NASH [23], even in the absence of cirrhosis in some cases but appears to occur less frequently than in hepatitis C virus-related liver disease [22]. Several small case series in children document that fibrosis progression can occur even during childhood, with progression to cirrhosis occurring in as short a time span as two years in some cases [24]. In adult series, potential clinical predictors of progression include more pronounced insulin resistance, progressive weight gain, age, and inflammation. Prognostic indicators in childhood remain unknown because of the small sample size of reported cases. NASH-related cirrhosis has increased as the primary indication for liver transplantation in young adults. Among 330 patients less than 40 years of age undergoing liver transplantation in the USA for NASH-related cirrhosis from 1987–2012, 4% were less than 18 years of age and 16% were 18–29 years old [9].

The overall impact of NAFLD on mortality remains unclear, with inconsistent results in adults and lack of any data in children. Initial studies in regional referral cohorts of adults suggested that NAFLD conferred a higher risk for overall mortality and liver-related mortality compared with reference general populations. In a community-based Minnesota study of 420 adults diagnosed with NAFLD between 1980 and 2000, predominantly by imaging studies, overall survival was lower than expected compared with the general population, with a standardized mortality ratio of 1.34 (95% CI, 1.003–1.760; p = 0.03) [5]. Liver disease was the third leading cause of death, occurring in 13% of those who died, compared with the thirteenth leading cause in the general population. The two leading causes of death were malignancy and ischemic heart disease (28% and 25% of the deceased, respectively). Independent risk factors for death were age, impaired fasting glucose, and cirrhosis, but the authors did not adjust for confounders despite the strong association of NAFLD with insulin resistance, central obesity, diabetes, and lipid abnormalities.

In a study of 131 adults with biopsy-proven NAFLD from the Cleveland Clinic, 78 (59.5%) died over a median follow-up period of 18.5 years, with maximum follow-up of 28.5 years. The three most common causes of death were coronary artery disease, malignancy, and liver-related death. Those with NASH had a significantly higher liver-related mortality (17.5%) than those with non-NASH NAFLD (2.7%, p = 0.005). Independent predictors of liver-related death were histologic NASH at time of initial biopsy, diabetes mellitus type 2, older age at biopsy, lower serum albumin, and increased serum levels of alkaline phosphatase (all p < 0.05).Independent predictors of overall mortality were type 2 diabetes, older age at biopsy, lower serum albumin, and higher serum glucose.

While these regional cohort studies raise concern for a heightened risk of death in patients with NAFLD, and for those with NASH, three of four national cohort studies in subjects drawn from NHANES III and using recommended stratified analysis methodology did not find an overall increased risk of all-cause mortality associated with NAFLD. Different cut-off values of abnormal ALT were used to define NAFLD in the studies and ultrasound was also used to define steatosis in a third, which accounts in part for different numbers of subjects. An analysis of 980 patients with suspected NAFLD and 6,594 patients without NAFLD, with a mean of 8.7 years of follow-up, found no statistically significant increase in overall mortality for the entire cohort with NAFLD [25]. However, in a subgroup of patients aged 45–54 years, suspected NAFLD was associated with higher all-cause mortality (hazard ratio (HR), 4.40; 95% CI, 1.27–13.23) and cardiovascular mortality (HR, 8.15; 95% CI, 2.0–33.2), after adjusting for conventional cardiovascular disease risk factors. No statistically significant increase in overall mortality was associated with elevated serum ALT in overweight, although elevated ALT was associated with increased liver-related death [26]. A more recent analysis drawn from NHANES III, and stratifying patients into those with and without sonographic evidence of steatosis and those with and without increased liver enzymes, also found no statistically significant increased risk of death from all causes, cardiovascular disease, or liver disease. The lack of an impact of NAFLD on overall mortality has also been supported by a large cohort study of patients in Denmark. Limitations of all of these studies include a relatively short period of follow-up from time of initial diagnosis, and lack of biopsy data to differentiate between NASH and non-NASH NAFLD.

Even though the presence of presumed NAFLD does not appear to confer an increased risk of overall mortality, it remains clearly associated with features of the metabolic syndrome and may independently confer a higher risk of cardiovascular disease, including increased carotid intima media thickness in adults. Ominously, children with NAFLD also have a significantly higher prevalence of the metabolic syndrome than obese children without evidence of NAFLD (50% and 15%, respectively; p < 0.001). In 100 obese children with NAFLD confirmed by ultrasound and elevated liver enzymes with exclusion of other causes, flow-mediated dilation of the brachial artery was significantly impaired and carotid intima media thickness significantly greater than in 150 obese children without evidence of NAFLD [27]. Children with NAFLD had 2.25 times the risk of low percentage flow-mediated dilatation and two times the risk of increased carotid intima media thickness, after adjusting for age, gender, Tanner stage, and presence of metabolic syndrome [27]. The independent associations of carotid intima media thickness with NAFLD, BMI, waist circumference, and systolic blood pressure (all p ≤ 0.005) were also confirmed in a population-based study of 642 adolescents aged 11–13 years in southern Italy [28]. The increased presence of these preclinical markers of atherosclerosis suggests that these children are likely to face high rates of cardiovascular events in the future, in accordance with adult data.

Pathogenesis

Given the global disease burden and accumulating epidemiological data since the early 1990s, the pathophysiology of NAFLD has been extensively investigated in both animal and human studies. While the factors that have fueled this epidemic, including our increasingly sedentary lifestyle and easy and cheaply available calorie-dense processed foods, are clear, it is the disproportionate morbidity across various clinical subgroups, races, and ethnicities that has energized moves to improve understanding of the pathogenic risk factors involved in NAFLD progression. It is now clear that not all NAFLD is created equal. While steatosis alone has been associated with insulin resistance and type 2 diabetes, the presence of NASH at diagnosis has been linked to a higher risk of hastened progression, end-stage liver disease, and need for liver transplantation. Therefore, the primary focus of recent research into pathogenesis of NASH has been to distinguish between the various constituents of the NAFLD spectrum, including steatosis alone and more advanced NASH, and then to understand the key drivers of these differences.

Lipids, Lipotoxicity, and the Non-Alcoholic Fatty Liver Disease

Puri et al. [29] first performed lipidomic analysis of the NAFLD spectrum in adults and reported that the mean triacylglycerol to diacylglycerol ratios were higher in NASH than in steatotic or normal livers (31, 26, and 7, respectively). There was also a similar stepwise increment in hepatic free cholesterol, which was significantly increased in NASH, while the free cholesterol to total phosphatidylcholine ratio increased progressively (0.34, 0.69, and 0.71, respectively) [29]. A follow-up study of plasma lipodomic profiles in adults revealed a stepwise increase in the lipoxygenase “HEPE” metabolites 5-(S)-, 8(S)-, and 15(S)-hydroxyeicosatetraenoic acid that characterized a progression from normal to steatosis to NASH. The level of 11(S)-hydroxyeicosatetraenoic acid, a non-enzymatic oxidation product of arachidonic (20:4) acid, was significantly increased in NASH alone. Parallel lipidomic analysis of murine and human liver tissues determined that mice maintained on a high-fat or high-fat high-carbohydrate diet provide a reproducible model of progressive liver disease. These studies demonstrated that lipid species may serve as markers of advanced liver disease and, importantly, marked increases in diacylglycerol species and oxidative stress markers such as oxidized coenzyme Q could be hallmarks of NASH with fibrosis [30]. There is now growing direct lipidomic data in children as well, suggesting increased total cholesterol, low density lipoprotein (LDL) cholesterol, and triglycerides in children with the NAFLD spectrum. There is also a correlation between worsening atherogenic profile and biopsy-proven NASH when compared with steatosis alone [31]. Similarly, a pilot study linked the increased intake of fructose to plasma oxidized LDL levels in children. The impact of dietary constituents on the NAFLD spectrum disorders is discussed in more detail below. These data together lead to the conclusion that oxidative stress-related liver injury, possibly through increased diacylglycerol species, is involved in progression of isolated steatosis to NASH.

Mitochondrial and Endoplasmic Reticulum Stress

Mitochondrial damage is well described in human NAFLD. Sanyal et al. [32] reported that there was loss of mitochondrial cristae and paracrystalline inclusions in nine of ten subjects with NASH, compared with zero of six subjects with steatosis alone. Terminal deoxynucleotidyl transferase (dUTP) nick end labeling (TUNEL) detects DNA fragmentation; TUNEL-positive cells denoting cell death from apoptosis, potentially from mitochondrial injury, were significantly increased in liver biopsy specimens from patients with NASH compared with those with simple steatosis and controls. Unexpectedly, TUNEL-positive cells were also greater in patients with NASH than in those with alcoholic hepatitis. Immunohistochemistry demonstrated active caspases 3 and 7 in NASH specimens, confirming the occurrence of apoptosis in this disease [33]. Further, this ongoing mitochondrial damage may be associated with reactive oxygen species release and progression of disease to more severe forms [34]. Endoplasmic reticulum stress associated with the unfolded protein response has also been shown to be more characteristic of liver biopsies from patients with NASH compared with those with normal liver histology or steatosis alone. Specifically, mRNA expression for C/EBP homologous protein (CHOP) and protein-content activated Janus-N kinase (known endoplasmic reticulum stress downstream elements) were increased in patients with NASH (Figure 36.1).

Figure 36.1 Pathogenesis of NAFLD-NASH.

Weight gain and obesity result in adipose tissue expansion and infiltration by macrophages with development of local insulin resistance and release of free fatty acids (FFA) into circulation. These events are thought to be critical for the development of hepatic steatosis which in turn contributes to local hepatic insulin resistance. Early in the disease process, the liver adapts to excess FFA; however, overtime, these adaptive mechanisms fail, resulting to lipotoxicity. Lipotoxicity can trigger multiple deleterious pathways in hepatocytes including ER stress, upregulation of death receptors, and mitochondrial dysfunction. Therefore, there is increased production of a wide variety of reactive oxygen, nitrogen, and lipid species, which dysregulate multiple redox sensitive signaling pathways leading to further increase in TG accumulation. The ensuing disruption of mitochondrial function triggers caspase activation and cell death; key “hits” in the progression to NASH. In conjunction, release of reactive species from hepatocytes can trigger activation of immune cells and convert the normally quiescent stellate cells to a fibrosis-inducing cell. This combination of events that are critical for the progression from steatosis to steatohepatitis to fibrosis is present to different extents in the various animal models that were more commonly used to study NAFLD and NASH. Abbreviations: AT, adipose tissue; HSC, hepatic stellate cells; ROS, reactive oxygen species; ER, endoplasmic reticulum; a-SMA, alpha-smooth muscle actin.

Reproduced with permission from Journal of Hepatology Kohli et al., 2011; 55 (941–3).

Genetic Heterogeneity and Risk

Multiple genome-wide associations and other genetic studies of adults have now validated the impact of genetic polymorphisms on disease severity and even heritability within the NAFLD spectrum. The gene that has the strongest association with NAFLD progression is PNPLA3, encoding the patatin-like phospholipase domain-containing protein 3 (acylglycerol O-acyltransferase). There is mounting evidence that polymorphisms in this gene modify the natural history of NAFLD in adults. Two recent meta-analyses concluded that the effect of the GG (I148M; rs738409) allele was to increase fat accumulation in the liver. Homozygosity in GG increased necro-inflammatory histological scores (3.24-fold higher) and risk of developing fibrosis (3.2-fold higher) when compared with CC homozygous individuals. Further, the same rs738409 GG genotype vs. the CC genotype was associated with a 28% increase in serum ALT [35]. These disease-modifying PNPLA3 polymorphisms are most frequently found in the Hispanic population, a group at higher risk of NASH, and are significantly less common within the African-American population, who appear to be relatively protected from hepatic steatosis and NASH. These adult PNPLA3 data have now also been reproduced in pediatric cohorts. The prevalence of the G allele was found to be higher in a multi-ethnic cohort of obese children and adolescents with hepatic steatosis, as determined by MRI [36]. Another study from Italy, with 149 children with biopsy-proven NAFLD (6 to 13 years of age), reported the PNPLA3 rs738409 polymorphism to be associated with severity of steatosis, hepatocellular ballooning, lobular inflammation, and perivenular fibrosis [37].

Another gene that has been studied as a modifier of NAFLD progression is APOC3, encoding apolipoprotein C3. In a cohort of healthy young Asian-Indian men, carriers of the APOC3 variant alleles (C482T, T455C, or both) had a 38% rate of NAFLD, even though their mean BMI was <25, while none of the wild-type homozygotes had NAFLD [38]. The data regarding the influence of APOC3 polymorphisms are not that clear in children. A study of 455 obese children failed to find any association between the APOC3 gene variants and hepatic steatosis [36].

Similarly, genes regulating oxidative stress pathways have been implicated in studies in adult and pediatric NAFLD. A common non-synonymous polymorphism in SOD2, encoding an antioxidant superoxide dismutase 2 (C47T; rs4880) was found to be associated with more advanced fibrosis in NASH with an odds ratio of 1.56 [39]. In a study of 234 obese Taiwanese children, a variant (UGT1A1*6) in UGT1A1, encoding a UDP-glucuronosyltransferase, is associated with higher relative rates of plasma bilirubin levels and was protective against NAFLD [40].

In addition, novel genetic variants, distinct from those identified in adults, have also been identified in a cohort of 234 Hispanic boys with histologically confirmed NAFLD, including the non-alcoholic fatty liver disease activity score with a SNP on chromosome 8 in the TRAPPC9 region and fibrosis stage with a SNP on chromosome 20, near action related protein 5 homolog. However, further replication studies will be needed to determine the clinical significance in other pediatric cohorts. Notably, in that study, a minority of the identified SNPs were previously reported in prior studies in adult patients as associated with NAFLD histological activity (two of 26 SNPs) or fibrosis grade (five of 26 SNPs). However, variants encoding PNPLA3 and TM6SF2 did emerge in association with fibrosis, as previously reported in adult genome-wide association studies in adults [41].

Nutritional Factors

It is self-evident that nutritional intake influences the prevalence and outcomes of NAFLD, a disease closely linked to excess caloric intake and adiposity. As discussed above, there are genetic factors that can predispose an individual or particular ethnic group to worsening insults from similar or fewer nutritional mal-influences. It is, therefore, important to understand these nutrient insults and their mechanism of liver injury in the NAFLD spectrum.

A nutritional constituent that has received the most attention and research is fructose. An analysis of BMI, gender, calorie intake, and age-matched patients from the NASH clinical research network database (https://jhuccs1.us/nash/) indicated that increased fructose consumption was associated with lower steatosis grade and higher fibrosis stage [42]. A further study of children using the same database showed that uric acid, a surrogate for total fructose consumption, was significantly associated with biopsy-proven NASH [43]. Interestingly, this study did not report significant differences in overall energy consumed from fat, carbohydrates, and protein in children with steatosis alone or NASH. Therefore, fructose consumption itself seems to be an independent predictor of more severe disease (NASH) within the NAFLD spectrum. A recent intervention study showing improvement in NAFLD with targeted reduction in sugar intake in children has also bolstered the data behind the role of fructose in the pathogenesis of NASH in children [44].

Other than fructose, sugar-sweetened beverages as a whole have been linked to obesity in general and metabolic syndrome in particular through database studies such as NHANES [43]. Natural compounds have also been thought of as treatment options for NAFLD, as discussed below. Dietary intake also modulates the intestinal microbiome, which is increasingly recognized as a factor influencing the severity of NAFLD in both children and adults. Children with biopsy-confirmed NAFLD and NASH have been shown to have intestinal dysbiosis, with progressively lower α-diversity in fecal microbiome samples, compared to control children without liver disease 3.52, NAFLD 3.36, borderline NASH 3.37 and NASH 2.97, respectively p = 0.001 [45].

Nature Versus Nurture

Collectively, studies have increasingly implicated both genetic and environmental influences on the progression and severity of NAFLD spectrum disorders. Initially a “two-hit” hypothesis had been proposed wherein the presence of steatosis was the first hit and then a “hypothesized” second hit would promote progression of the liver injury. This hypothesis has since been revised to now consider a more nuanced “multiple hit” hypothesis, which includes a greater understanding of the role of “nurture,” including diverse environmental factors, interacting with “nature,” or genetic susceptibilities which predispose to greater lipotoxicity when exposed to these deleterious environmental influences. For example, an exciting new area of research is now emerging on the potential contribution of pervasive endocrine-disrupting environmental toxicants, such as perfluoroalkyl substances or arsenic, in predisposing to increased prevalence of NAFLD in humans [46, 47]. Thus, a complex interplay of environmental and genetic factors is likely to account for the conundrum as to why certain individuals with hepatic steatosis are protected from progressing to severe liver disease compared with others who are more predisposed to disease progression, fibrosis, cirrhosis, and liver-related morbidity.

Clinical Diagnostic and Prognostic Tools

Clinically, the spectrum of NAFLD remains challenging to diagnose and monitor as it is often clinically silent with few overt signs and symptoms in children, even in patients with NASH. The most commonly reported symptoms in children are abdominal pain and fatigue. In some cases, a complaint of abdominal pain prompts imaging studies that identify abnormal hepatic steatosis and lead to referral. The most common physical signs include obesity, in particular, central adiposity, hepatomegaly, acanthosis nigricans, and splenomegaly. However, NAFLD can also present in thin individuals with severe insulin resistance caused by lipodystrophy or in individuals who have centralized obesity but an overall normal BMI. In childhood, NAFLD presenting with overt jaundice and signs of end-stage liver disease is very rare, with one case report of a Hispanic girl aged 11 years who had grade 3 esophageal varices and cirrhotic NASH [48]. She subsequently underwent liver transplantation at age 20 years, which to our knowledge is the youngest reported case of the use of transplantation for NASH diagnosed in childhood. There have also been case reports of two children who developed cirrhosis at ages ten and 14, with one child progressing to portal hypertension and variceal bleeding within two years of diagnosis.

Many patients with NAFLD are first identified by elevated ALT or AST found incidentally or during screening of an overweight and obese child for comorbid conditions. The American Academy of Pediatrics Expert Committee Recommendations Regarding the Prevention, Assessment and Treatment of Child and Adolescent Overweight and Obesity suggests that, in the current absence of evidence-based recommendations for screening, biannual screening of ALT and AST should begin at age ten for children with BMI 95th percentile and in children with BMI of 85–94th percentile if other cardiovascular disease risk factors are present [49].

Associated medical conditions that appear to confer a higher risk of NAFLD include insulin resistance and diabetes mellitus type 2, polycystic ovarian syndrome, hypertriglyceridemia, hypothalamic obesity, lipodystrophy, and possibly obstructive sleep apnea. A thorough screening should be done in any patient diagnosed with NAFLD or NASH, as many children may be unaware that they have such concomitant illnesses as:

• 
  

  Viral hepatitis: hepatitis C

  
  
• 
  

  Autoimmune hepatitis

  
  
• 
  

  Storage diseases:

  
  
  
  
  • 
    

    Wilson disease

    
    
  • 
    

    Hemochromatosis

  
  
  
  
• 
  

  Metabolic diseases:

  
  
  
  
  • 
    

    glycogen storage diseases

    
    
  • 
    

    abetalipoproteinemia and hypobetalipoproteinemia

    
    
  • 
    

    lipodystrophy

    
    
  • 
    

    cystic fibrosis

    
    
  • 
    

    urea cycle disorders

    
    
  • 
    

    fatty acid transport and oxidation disorders

    
    
  • 
    

    carnitine deficiency

    
    
  • 
    

    oxidative phosphorylation disorders

    
    
  • 
    

    lysosomal acid lipase deficiency

  
  
  
  
• 
  

  Medication toxicity:

  
  
  
  
  • 
    

    glucocorticoids high-dose estrogen valproic acid amiodarone

    
    
  • 
    

    aspirin-induced Reye syndrome

    
    
  • 
    

    parenteral nutrition

    
    
  • 
    

    collagen vascular disorders: juvenile rheumatoid arthritis (often medication induced)

  
  
  
  
• 
  

  Adverse nutritional intake:

  
  
  
  
  • 
    

    alcohol abuse

    
    
  • 
    

    malnutrition

    
    
  • 
    

    starvation

    
    
  • 
    

    kwashiorkor

    
    
  • 
    

    malabsorptive disorders.

  
  

Identification often requires further specific therapy for these conditions.

Anthropometrics

The presence of metabolic syndrome is quite high in children with NAFLD, up to a 50–66% prevalence found in large single center cohort studies in both the USA and Italy. Metabolic syndrome has been shown to be associated with central adiposity, and waist circumference has been found to be predictive of worsening and progression of obesity comorbidities. A study of 201 obese children in the context of NAFLD spectrum disorders found that increased waist–hip ratio was associated with elevated serum aminotransferases [50]. Higher metabolic and cardiovascular risks are associated with higher waist-to-height ratios in obese children. The strongest evidence to date linking anthropometric measures and histological severity of NAFLD came from the work of Manco et al. [20], who have reported data from a study of 197 Caucasian children 3 to 19 years of age showing that increasing waist circumference correlated with biopsy-proven liver fibrosis. Nobili et al. [51] validated a combination of non-invasive markers and reported them collectively as the Pediatric NAFLD fibrosis index (PNFI) in 141 children with fibrosis on liver biopsy. They proposed a final model based on age, waist circumference, and triglycerides, in which a PNFI 9 had an area under the receiver operating characteristic (ROC) curve of 0.85 for the prediction of liver fibrosis, within their cohort (positive likelihood ratio, 28.6). Therefore, anthropometric measures may be useful as clinical discriminators individually or as part of expanded diagnostic panels to discriminate need for liver biopsy. However, further validation of such diagnostic panels is needed.

Routine Laboratory Tests

Routine hepatic laboratory measures (serum aminotransferase levels) have been used as primary screening tools to identify obese individuals at risk of NAFLD. Unfortunately, it became clear early on that these measures alone are insensitive in identifying NAFLD spectrum disorders in children and some children with NAFLD have normal liver enzymes. Therefore, serum ALT alone cannot be taken as a reliable indicator of NAFLD or disease severity. Further, repeated measures of ALT can fluctuate. Patton et al. [52] examined components of routine laboratory tests predictive of NAFLD pattern and fibrosis severity but found inadequate discriminate power to replace liver biopsy in evaluating pediatric NAFLD. In a prospective study of 176 children with biopsy–confirmed NAFLD, they determined that the area under the ROC curve model with AST and ALT was 0.75 (95% CI, 0.66–0.84) and 0.74 (95% CI, 0.63–0.85) for distinguishing steatosis from advanced NASH and bridging fibrosis from lesser degrees of fibrosis, respectively.

Finally, as mentioned earlier, there is no standard, outcome-based reference range for serum ALT that is universally accepted by all laboratories, such as exists for abnormal lipid values in adults. However, using a biologically determined threshold as low as 22 U/L in girls and 26 U/L in boys to detect liver disease and initiate further testing has not yet undergone any cost-effectiveness analyses and may generate a barrage of unnecessary and expensive tests. Therefore, in our clinical practice, we routinely initiate additional screening for liver disease if ALT elevations are persistently >40 U/L – roughly 1.5–2 times the upper limit of normal suggested by the NHANES analysis – on at least two measurements one month apart. Some studies have also suggested that elevated serum AST and gamma-glutamyltransferase (GGT) correlate better with fibrosis and/or improvement after weight loss, but the significance of isolated mild serum AST and GGT elevations in the absence of elevated ALT is less certain [53].

When persistent elevations in ALT are identified, and in some contexts AST and GGT, it is important to exclude other potential causes of chronic hepatitis, including viral hepatitis (B, C, and also A if indicated), autoimmune hepatitis, drug or alcohol injury, Wilson disease, hemochromatosis, and celiac disease through appropriate serological testing as well as by clinical history and physical examination. In summary, routine serum aminotransferases continue to guide practitioners in the evaluation of NAFLD but they cannot be used exclusively to screen and monitor progression.