Prevalence and presentation

Variation in reports of prevalence of NAFLD is partly due to the different methods of detection⁽Reference Musso, Gambino and Cassader^9, Reference Anderson, Howe and Jones^14, Reference Xanthakos, Jenkins and Kleiner¹⁵⁾. Although the gold standard for diagnosis is liver biopsy, this is not practical in population studies. One of the most veritable studies reporting prevalence in the general paediatric population is an autopsy study of CYP who died of accidental injury⁽Reference Schwimmer, Deutsch and Kahen¹⁶⁾. In this study, Schwimmer et al. reported a prevalence of NAFLD in 9·6 % of this cohort, with NASH (steatohepatitis) present in 3 %⁽Reference Schwimmer, Deutsch and Kahen¹⁶⁾. Other population studies in children vary in prevalence from 1·7 to 42·5 %⁽Reference Anderson, Howe and Jones¹⁴⁾. This variation is due to the different methods used to diagnose the condition and the different population studies both in terms of ethnicity and risk factors. In a study of morbidly obese children undergoing bariatric surgery, NAFLD was identified in biopsy in 83 %⁽Reference Xanthakos, Miles and Bucuvalas¹⁷⁾ in contrast to a longitudinal study of healthy children from the UK midlands using ultrasound, in which the prevalence was only 2·5 %⁽Reference Lawlor, Callaway and Macdonald-Wallis¹⁸⁾.

CYP often present with an incidental finding of elevated transaminases and an echogenic liver on ultrasound. CYP with NAFLD are likely to be overweight or obese but this is not universal⁽Reference Wattacheril and Sanyal¹¹⁾. Neither bloods nor imaging will differentiate those with significant fibrosis from those without⁽Reference Molleston, Schwimmer and Yates¹⁹⁾ and a liver biopsy is still required in some. Biopsy is not practical in most cases however, given its invasiveness, the need for sedation and the inherent risks involved. Both the European and the North American Societies for Paediatric Gastroenterology, Hepatology and Nutrition recommend biopsy in children in cases of clinically suspected advanced liver disease, in those in whom another disease is a possibility, before pharmacological treatment, and as part of a structured intervention protocol or clinical research trial⁽Reference Vos, Abrams and Barlow^20, Reference Vajro, Lenta and Socha²¹⁾.

All other possible causes of liver disease in children should be ruled out including liver-based metabolic disease⁽Reference Hegarty, Deheragoda and Fitzpatrick⁶⁾. A family history of metabolic syndrome should be sought in addition to medication use and other risk factors. A list of differential diagnoses is given in Table 1.

Table 1. Differential diagnosis of fatty liver in CYP

MDMA, 3,4-methyl endoxy metamphetamine; HCV, hepatitis C virus; Gal-1-PUT, galactose-1-phosphate uridyl transferase.

As liver biopsy is not always possible or felt appropriate, non-invasive methods are often used to determine the severity of disease in terms of fibrosis. Likewise, non-invasive methods of longitudinal monitoring are used as repeated biopsy is only rarely undertaken⁽Reference Fitzpatrick and Dhawan²²⁾. Although there is no consensus in terms of appropriate non-invasive measures, transient elastography is an imaging technique which is reliable in determining no fibrosis from significant fibrosis/cirrhosis⁽Reference Fitzpatrick, Quaglia and Vimalesvaran^23, Reference Nobili, Vizzutti and Arena²⁴⁾. MRI protocols can differentiate different degrees of steatosis and magnetic resonance elastography is a promising tool for fibrosis staging, however the cost is often preclusive⁽Reference Xanthakos, Podberesky and Serai²⁵⁾. There are blood-marker algorithms available which have been derived from adult populations, including the NAFLD fibrosis index and Fib4⁽Reference Musso, Gambino and Cassader^9, Reference Kaswala, Lai and Afdhal²⁶⁾, although none have been adequately validated in children. Algorithms which include age and markers of collagen turnover influenced by growth such as procollagen III peptide may not accurately reflect fibrosis in this population and rederivation of many of the available algorithms is needed. The paediatric NAFLD fibrosis index was developed using a discovery cohort of children and may be a useful tool, however external validation is required⁽Reference Alkhouri, Mansoor and Giammaria²⁷⁾.

Susceptibility to developing non-alcoholic fatty liver disease in childhood

Histology and differences in paediatric non-alcoholic fatty liver disease

The NASH Clinical Research Network in the USA convened a group of expert hepatopathologists to develop a scoring tool in NAFLD for use as an outcome measure in clinical trial⁽Reference Kleiner, Brunt and Van Natta⁶⁰⁾. This tool known as the NAFLD Activity Score separately assigns severity of steatosis (0–3), inflammation (0–3) and ballooning (0–2) to give a cumulative score. Fibrosis is scored separately by the NAFLD Activity Score with 0 indicating no fibrosis, F1 some pericentral fibrosis (1b is periportal fibrosis), through F2 and F3 (bridging fibrosis) to F4 nodular change or cirrhotic change. Although children were used in the initial discovery cohort, it has since become clear that the pattern of disease in many children is distinct and that the NAFLD Activity Score does not reflect these features.

CYP frequently though are found to have type-2 disease which is a more periportal tract based pattern⁽Reference Yeh and Brunt⁶¹⁾. Fifty to seventy percent have a type-2 pattern or a crossover between type 1 and type 2⁽Reference Fitzpatrick, Mitry and Quaglia^62–Reference Schwimmer, Behling and Newbury⁶⁴⁾. Type-2 disease has been studies in the context of severity, and both adults and children with histology more likely to have higher stage of fibrosis⁽Reference Brunt, Kleiner and Wilson^65, Reference Africa, Behling and Brunt⁶⁶⁾. It is not clear if this pattern is due to a separate pathophysiological mechanism, although it certainly seems to be a marker of more advanced NASH.

The pathophysiology of why the location of inflammation and fibrosis varies is not understood. One possibility which may in part explain the preferential distribution is the concept of zonation. Along the liver lobe, the hepatocytes have different functions depending on their location in the lobule. For example, periportal hepatocytes are functionally specialised in Kreb's cycle amongst other tasks and those in the area of the central vein are rich in cytochrome P450 enzymes. The exposure of the liver to dietary components is more marked in zone 1 than in zone 3 (pericentral)⁽Reference Nobili, Mosca and De Vito⁶⁷⁾.

It is likely that there is an evolutionary or a developmental reason for this localisation. It is possible that the repair mechanism in the liver, or rather the propensity and vigour with which the liver regenerates is different according to maturity and developmental stage. Fibroblast-specific protein 100, a fibroblast marker, has increased expression in interlobular ductal cells in paediatric control (healthy) liver than adult control liver, demonstrating that ductal epithelial cells in children have a greater tendency to exhibit features of mesenchymal cells and implying epithelial mesenchymal transition is more active within portal tracts in children⁽Reference Omenetti, Bass and Anders^68, Reference Omenetti, Choi and Michelotti⁶⁹⁾.

An elegant body of work by Anna Mae Diehl and colleagues describe the localisation of hedgehog (Hh) signalling in the periportal region⁽Reference Omenetti, Choi and Michelotti^69, Reference Swiderska-Syn, Suzuki and Guy⁷⁰⁾. Hh signalling is expected to be more active in children. The Hh pathway is a pivotal morphogenic signalling pathway central in organogenesis. The pathway becomes quiescent in the liver during adolescence and reactivates in the presence of injury. Thus Hh signalling is involved in the activation of the regenerative pathway of liver. Healthy paediatric livers demonstrate more Hh signalling than healthy adult livers. Increased exposure to Hh ligands stimulated cells involved in wound healing. Normally these cells are tightly regulated however when deregulated, chronic inflammation, fibrosis and liver cancer may result. If children possess a proportionally greater number of Hh ligand producing cells and Hh responsive cells, then children may be at particular risk from insults that promote activation of the Hh pathway⁽Reference Swiderska-Syn, Suzuki and Guy⁷⁰⁾. Guy et al. examined the association of portal fibrosis with the activation of the regenerative Hh signalling⁽Reference Guy, Suzuki and Abdelmalek⁷¹⁾. The number of Hh responsive cells correlated closely with both portal injury and the severity of fibrosis in liver biopsies of adult's patients with NAFLD (P < 0·0001) supporting a relationship between Hh activation and liver damage. For this study, immunohistochemistry for Hh-ligand and Gli1 (Hh responsive cells) was undertaken on ninety biopsies within the NASH Clinical Research Network repository. A follow-on study in children again demonstrated more Hh activation with good correlation of Hh ligand expression and Hh responsive cells to liver injury and severity of disease⁽Reference Swiderska-Syn, Suzuki and Guy⁷⁰⁾.

In contrast, adult livers have more active pericentral Wnt signalling which is involved in glutamine synthesis, drug metabolism, bile acid and haem synthesis, regeneration and the response to oxidative stress⁽Reference Gebhardt and Matz-Soja⁷²⁾.

Natural history and targeting intervention

The natural history of paediatric NAFLD has not yet been well described. Case series including one of 20 years describe occasional need for transplantation in young adulthood⁽Reference Gebhardt and Matz-Soja^72, Reference Molleston, White and Teckman⁷³⁾, but the rate of progression over years is not known⁽Reference Feldstein, Charatcharoenwitthaya and Treeprasertsuk⁷⁴⁾. Paired liver biopsies were analysed from 122 children who had enrolled in the placebo arm of two randomised clinical trials in NAFLD. Placebo groups were given lifestyle counselling for a period of 52 or 96 weeks. Over this time, fibrosis progressed in 23 % and improved in 34 %⁽Reference Xanthakos, Lavine and Yates⁷⁵⁾. In children who present in the pre-teenage years, already established with stage 2–3 fibrosis, the rate of progression may be accelerated⁽Reference Goh and McCullough⁷⁶⁾. The heterogeneity within the population is not yet well understood but variability in phenotype may be due to underlying genetic susceptibility rather than environmental exposure.

The relative histological severity at presentation in children with this disease and the fact that alcohol is an unlikely confounding factor, means that paediatric NAFLD serves as an excellent disease model in evaluating pathophysiological mechanisms of development and thus targeting intervention in predisposed individuals.

Lifestyle change resulting in weight loss is an effective way of reversing or stabilising disease. In a meta-analysis of adults with NAFLD, weight loss of 5 % or more resulting in improvement in steatosis whereas ≥7 % weight loss resulted in improvement in steatohepatitis and in those with ≥10 % weight loss, all features of NAFLD were reversed or stabilised⁽Reference Musso, Cassader and Rosina⁷⁷⁾. In a prospective study again in adults these outcomes were confirmed⁽Reference Vilar-Gomez, Martinez-Perez and Calzadilla-Bertot⁷⁸⁾. Only 50 % of the cohort were able to achieve 7 % of weight loss or more though of note in 94 % of those who achieved ≥5 % weight loss, fibrosis stabilised or reversed.

A small number of trials in children have demonstrated similar results. In an Italian study of eighty-four children, weight loss (average 4 kg) over a 12 months period achieved an improvement in alanine aminotransferase and steatosis on ultrasound⁽Reference Nobili, Marcellini and Devito⁶³⁾. Of the eighty-four children, fifty-seven (70 %) children completed the 12 month intervention with a mean 8 (sd 4·7) % decrease in weight in the fifty two who were overweight or obese. In the remaining five children who completed the study and had a BMI <85th centile, weight remained unchanged but alanine aminotransferase levels improved in two and normalised in three patients. Another paediatric study of intensive lifestyle intervention in North America achieved improvement in BMI z-score with a decrease of 0·1 U (P < 0·05) baseline to 1 year and decrease in alanine aminotransferase in 69 % of the follow-up cohort. There was a 53 % drop-out rate however⁽Reference DeVore, Kohli and Lake⁷⁹⁾.

Insulin resistance is well recognised as an accompanying feature in 70 % of those with NAFLD, although not clearly associated with more or less severe disease in terms of inflammation and fibrosis. Insulin sensitisers have been studied frequently in clinical trial but without a consensus as to their effectiveness. In adults, the PIVENS trial was a randomised controlled trial of pioglitazone, vitamin E and placebo in treatment of adults with NASH but without type-2 diabetes over 96 weeks⁽Reference Chalasani, Sanyal and Kowdley⁸⁰⁾. The specified outcome measure of improvement in the NAFLD Activity Score by ≥2 points was not reached with significance using pioglitazone. In children the TONIC trial compared metformin, vitamin E and placebo in 173 children with biopsy proven NAFLD⁽Reference Lavine, Schwimmer and Van Natta⁸¹⁾. Again, there was no statistically significant difference in the outcome measure (improvement in alanine aminotransferase) in those treated with metformin compared to those with placebo. Interestingly a change in homoeostatic model assessment-insulin resistance was not seen in those treated with metformin suggesting perhaps that treatment dose may have been insufficient or that selecting out those with insulin resistance may be more appropriate. The likelihood is that as the disease is relatively heterogeneous, treatments may need to be individually tailored.

Oxidative stress, most likely mediated by accumulation of fat droplets and the low grade inflammatory response accompanying visceral adiposity in the setting of genetic predisposition, is known to occur and perpetuate injury in NAFLD. Antioxidants, most commonly vitamin E have been trialled. Vitamin E has been shown in PIVENS to reduce steatohepatitis⁽Reference Sanyal, Chalasani and Kowdley⁸²⁾ and in TONIC to reduce ballooning. A significant difference in the main outcome measure (alanine aminotransferase) in the paediatric study TONIC was not found however⁽Reference Lavine, Schwimmer and Van Natta⁸¹⁾.

Current recommendation in adult practice is to use vitamin E in non-diabetic patients with biopsy proven NASH⁽Reference Chalasani, Younossi and Lavine⁸³⁾. The PIVENS study has not yet been validated adequately however and thus caution is advised generally. There is no consensus on vitamin E use in children with NAFLD.

Ursodeoxycholic acid is another antioxidant which has been used in trial though without consistent effects. Cysteamine bitartrate works by increasing glutathione synthesis. This was used in the CyNCh trial in children with NAFLD in comparison to placebo over 52 weeks. Although there was an improvement in alanine aminotransferase, the primary outcome of histological improvement was not achieved with statistical significance⁽Reference Schwimmer, Lavine and Wilson⁸⁴⁾.

The gut microbiome and its influence on bile acid metabolism is a new and emerging area in NAFLD and one which has led to the development of a cluster of new potential therapeutic agents. It is now recognised that the consumption of obesogenic foods leads to a change in the microbiota of the gut and to an increased permeability of the intestinal epithelium⁽Reference Kalliomaki, Collado and Salminen⁵¹⁾. The microbiota influence the development of NAFLD through several different mechanisms among which is the production of SCFA which stimulate de novo triglyceride synthesis in the liver, modulation of choline metabolism (involved in VLDL synthesis), lipopolysaccharide production and of course modulation of bile acid metabolism⁽Reference Chen, Thomsen and Vitetta⁸⁵⁾.

The association between obesity and gut microbial change is well established in the form of a lower diversity of microbiota, although it is not yet clear whether this is a causal or a secondary effect. The change in microbial species leads to a higher concentration of secondary bile acids in the enterohepatic recirculation. In healthy individuals, bile acids play an important role in glucose and lipid metabolism regulating the negative feedback loops. Bile acids act via cellular receptors including farnesoid X receptor and G protein-coupled bile acid receptor to affect metabolism⁽Reference Chen, Thomsen and Vitetta^85, Reference Kuipers, Bloks and Groen⁸⁶⁾.

Probiotics and prebiotics are an alternative way to attempt to re-establish a healthy diversity of flora however there is no clear guidance on specific formulation or dose⁽Reference Chen, Thomsen and Vitetta⁸⁵⁾. Weight loss, whether dietary induced or following bariatric surgery has been found to have a similar effect. Other important ways to disrupt the enterohepatic recirculation of bile thus decreasing the total bile pool is by using farnesoid X receptor agonists⁽Reference Neuschwander-Tetri, Loomba and Sanyal⁸⁷⁾.

The process of fibrosis is in itself the common end point of several processes; steatosis with oxidative stress, inflammation and apoptosis. Prevention of fibrosis is optimal although antifibrotics are entering the clinical trial arena.

Although there is major interest in the search for the magic bullet for NAFLD, in reality the heterogeneity of the condition means that individualised treatment is needed, for example a patient with markers of increased oxidative stress may respond better to an antioxidant whereas those who have a poorly diverse microbiome may respond better to its modulation.

Everything considered, the most effective treatment (reversal) of NASH/steato-fibrosis is weight loss.

As discussed, there is evidence to show that interventions resulting in 5–10 % weight loss results in improvement in NASH⁽Reference Musso, Cassader and Rosina⁷⁷⁾. In >50% however, this, weight loss is not achieved. The barriers to this weight loss need to be addressed. In children, weight loss per se may not be necessary as they grow into their centile. Thus, it is difficult to be prescriptive regarding the quantity of weight that needs to be lost and often CYP may require a BMI z-score to suit their specific genetic predisposition. Interestingly in a recent study of lifestyle intervention, there was a greater improvement in steatosis with weight loss in those with a higher genetic risk profile⁽Reference Ma, Hennein and Liu⁸⁸⁾.

The importance of recognising mental health

Why does lifestyle intervention not work for all? Adherence may be the principal challenge and given that CYP are almost entirely dependent on family being involved in their goal, this increases complexity of the problem.

Social deprivation, a lack of education about the importance of a healthy lifestyle and cost of fresh, unprocessed foods are all important considerations and not all which can be successfully managed in the context of a clinical lifestyle intervention programme.

An underexplored area in ability to undertake and maintain lifestyle change in this population is the importance of mental health. The prevalence of depression and anxiety in CYP and adults with obesity is significant⁽Reference Leon, de Klerk and Ho^89, Reference Mannan, Mamun and Doi⁹⁰⁾. Depression is over-represented in adults with NAFLD as demonstrated in survey of 567 patients as part of the NASH Clinical Research Network with 53 % exhibiting subclinical depression and 14 % clinical depression, 45 % with subclinical anxiety with 25 % clinical anxiety. There were some histological correlates with presence of depression⁽Reference Youssef, Abdelmalek and Binks⁹¹⁾. A small number of studies have shown that quality of life in both adults and children with NAFLD are significantly inferior to normal controls. In a survey of 239 children apart of the NASH Clinical Research Network, children with NAFLD had worse total physical and psychological quality of life scores as determined by the PedsQL questionnaire. Fatigue, trouble sleeping and sadness accounted for almost half the variance in quality of life scores⁽Reference Kistler, Molleston and Unalp⁹²⁾. In adults, degree of tiredness reported by patients with NAFLD was similar to that in women with primary biliary cirrhosis, a condition in which fatigue is often profound⁽Reference Newton, Jones and Henderson⁹³⁾. Although obstructive sleep apnoea may contribute to daytime somnolence in obesity, non-obstructive sleep apnoea sleep problems and the pathogenesis of this fatigue has not yet been systematically studied in NAFLD.

Response to lifestyle intervention in NAFLD is significantly less effective in the presence of depression particularly in the presence of acute major depressive disorder⁽Reference Tomeno, Kawashima and Yoneda⁹⁴⁾. Although some studies in adults have used a psychological approach to lifestyle change in obesity related disorders, no trials in children are reported⁽Reference Gelli, Tarocchi and Abenavoli^95–Reference Bellentani, Dalle Grave and Suppini⁹⁷⁾. This approach included counselling sessions and cognitive behavioural therapy.

Both adult and paediatric studies have drawn a link between depression, inflammation and the coexistence of complications of obesity, most frequently visceral adiposity associated conditions⁽Reference Kiecolt-Glaser, Derry and Fagundes^98–Reference Byrne, O'Brien-Simpson and Mitchell¹⁰⁰⁾. When combined with predisposing factors, immune challenges can lead to exaggerated or prolonged inflammatory responses⁽Reference Kiecolt-Glaser, Derry and Fagundes⁹⁸⁾. The resulting sickness behaviours, depressive symptoms and poor lifestyle choices may lead to further inflammation. This pattern suggests that effective treatment options which target this vicious cycle may halt both the amplified inflammation and depressive symptoms.