Liver fibrosis in children: a comprehensive review of mechanisms, diagnosis, and therapy

1) Biliary atresia

BA is a progressive fibroinflammatory disease involving a segment or the entire extrahepatic biliary tree that leads to obliteration of the affected bile ducts within the first 3 months of life. To date, the exact pathophysiology of BA has remained elusive with a focus on environmental insults, immunologic dysregulation, genetic susceptibility, and developmental defects. The current hypothesis endorses a viral insult, leading to overactivation of the immune system. Fibrosis in BA is unique in that it progresses rapidly, leading to cirrhosis within a few weeks of birth in untreated cases. Kasai portoenterostomy (KPE) bypasses the atretic extrahepatic bile ducts and restores bile flow. The therapeutic outcome of KPE was evaluated using serum bilirubin levels at 6 months postoperative. Despite a good clinical response to KPE, liver fibrosis in BA often progresses to cirrhosis, requiring LT. In different series published worldwide, the rate of needing LT after KPE was 45%–70% at 5 years. Research has focused on identifying prognostic indicators in the post-KPE setting to predict progression to fibrosis and cirrhosis [9,10].

2) Primary sclerosing cholangitis

PSC is a chronic fibroinflammatory disorder of the intrahepatic and/or extrahepatic bile ducts of unknown etiology. The development of fibrosis is significant in pediatric PSC [11-16]. Despite the inconspicuous symptoms at presentation, such as fatigue and favorable magnetic resonance cholangiopancreatography scores, the majority of pediatric patients have significant fibrosis in the initial liver biopsy [17,18]. Around 1%–2% of adults, but 17%–30% of children progress to advanced liver disease and require LT each year [12,19]. In a large multicenter study of 781 children, Deneau et al. [11] found that a high fibrosis scoring aspartate transaminase (AST) to platelet ratio index (APRI) portends a poor prognosis.

No treatment options to date, such as oral vancomycin or ursodeoxycholic acid, are associated with improved fibrosis in pediatric PSC [20,21]. Haisma et al. [22] performed a genome-wide association study of 29 patient-parent trios with PSC onset at ≤12 years of age. They found pathogenic variants in 22 of 29 of patients and that some of the genes were involved in transmembrane transport, adaptive and innate immunity, and epithelial barrier function. On the other hand, adult genome studies identified most adult-onset susceptibility genes to the human leukocyte antigen complex II locus [23,24]. Therefore, pediatric PSC may represent a heterogeneous disease characterized by differing mechanisms and potential for fibrosis that may benefit from individualized treatment.

3) Alagille syndrome

Alagille syndrome is a multisystem disorder of morphogenesis that characteristically involves cholestasis, heart murmur, structural heart disease, dysmorphic facial features, vertebral anomalies, and optical findings. Through a mutation in the JAG1 gene on chromosome 20, commonly NOTCH2, the Notch signaling pathway is perturbed, resulting in complex developmental manifestations [25]. Liver involvement in Alagille syndrome is characterized by bile duct paucity, cholestasis, and fibrosis. Although the liver disease may improve with age, 43.5%–76% of patients reportedly require LT by adulthood [26-28].

Defective Notch signaling associated with Alagille syndrome has guided our understanding of biliary fibrosis. Upon injury, hepatic progenitor cells (HPCs) are activated and have the capacity to differentiate into hepatocyte as well as biliary epithelial cells in hepatocellular and biliary injury, respectively. Jagged-1 expressing portal MFs mediate the differentiation of HPCs upstream of Notch signaling. The outcome of Alagille syndrome is a scarcity of reactive ductular cells, resulting in thinner portal septa and an abundance of intermediate hepatic-biliary cells that promote intralobular chicken wire fibrosis [29,30].

1) Chronic viral hepatitis

Most children acquire hepatitis B virus (HBV) and/or hepatitis C virus (HCV) infections in infancy, with 90% and 60% developing chronic infections, respectively [31,32]. Overall, chronic HCV infection in children follows an indolent course with low-level fibrosis. In 2 cohorts of 121 and 332 children, only 2 and 6 children developed cirrhosis in longitudinal follow-up, respectively [33,34]. Despite vaccination, HPV remains an important source of mortality and morbidity in children in the United States. In a study of histological specimens collected from 134 children, fibrosis and cirrhosis were present in 82% and 4%, respectively. In another study of 292 HBV-infected children, cirrhosis occurred in 3% at a median of 4 years of age [35,36].

In chronic HBV, Łotowska and Lebensztejn [37] found a positive correlation between the number of SMA-positive HSCs and fibrosis stage. The same group performed an ultrastructural phenotype analysis of chronic HBV biopsies and documented shifts in 2 cell types in children with advanced fibrosis: quiescent HSCs were replaced with transitional and MF-like HSCs and damaged liver progenitor cells in close proximity to the HSCs [38-40]. Close monitoring of fibrosis progression guides clinical decisions to begin antiviral therapy in these patients to prevent ongoing fibrosis [41].

2) Autoimmune hepatitis

Autoimmune hepatitis (AIH) is a chronic inflammatory disease of the liver characterized by hypergammaglobulinemia and autoantibody positivity. AIH is an important consideration in childhood liver fibrosis since one-third of patients present with cirrhosis and its complications [42,43]. Maia et al. [44] performed immunohistochemical examinations of pediatric AIH liver biopsies and noted that the activation of HSC populations decreased significantly following therapy. Studies involving follow-up biopsies in 16 and 13 pediatric patients with AIH demonstrated regression of fibrosis over 6–12 months [45,46].

1) NAFLD–nonalcoholic steatohepatitis

NAFLD is a disease spectrum that ranges from hepatocyte steatosis to nonalcoholic steatohepatitis (NASH) to fibrosis and ultimately cirrhosis [47]. Fibrosis in NAFLD-NASH is perhaps the most studied pathophysiology in terms of mechanisms, diagnostics, and therapeutics. Atsawarungruangkit et al. [48] performed transient elastography-based screenings of 740 US adolescents embedded within the 2017–2018 National Health and Nutrition Examination Survey and identified advanced fibrosis secondary to NAFLD in 2.84%. According to an evaluation of liver biopsies from 100 children with NAFLD, 2 phenotypes of pediatric NASH exist: type I, with lobular inflammation and perisinusoidal fibrosis (17% of biopsies); and type II, with portal inflammation and fibrosis (51% of biopsies) [49]. Notably, children with the type II phenotype had more significant fibrosis.

The factors involved in driving the progression of NAFLD to fibrosis have been extensively evaluated. The 2 consistent risk factors for fibrosis development are metabolic syndrome and waist circumference [50-52]. Genetic variation is another source of wide study. Recently, Tao et al. performed a meta-analysis of 42 studies (>1 million participants), including 10 pediatric studies, to assess the role of rs641738C>T in MBOAT7. Although the variant was associated with advanced fibrosis on biopsy, its functional implications remain unclear. Another widely established variant is the rs738409 polymorphism in PNPLA3/adiponutrin. A meta-analysis of 5 studies with histological fibrosis assessments showed a significant association between advanced fibrosis and the variant [53]. In a study of 54 patients with NAFLD, those carrying the PNPLA3 variant had more periportal/portal fibrosis, hepatic progenitor cell activation, and serum oxidative stress [54]. Whether these findings can be applied to other etiologies could pose a worthwhile research question.

Environmental influences, such as dysbiosis and in utero exposure to a maternal Western diet, are related not only to NAFLD, but also to early liver fibrosis. NASH et al. recently demonstrated that nonhuman primate fetuses exposed to a maternal Western-style diet displayed increased fibrillar collagen deposition in the liver periportal region with localized HSC and MF activation. In a similar model, Friedman et al. [55] prevented collagen fibril formation in the offspring of mice fed a Westernstyle diet through a potent antioxidant, pyrroloquinoline quinone, which reversed macrophage metabolic programming from oxidative to glycolytic pathways and improved gut tight junction expression. Therefore, dietary modulation may be an important intervention for all patients with liver diseases.

2) Intestinal failure-associated liver disease

IFALD encompasses a spectrum of liver diseases, including steatosis, cholestasis, and fibrosis, in patients with intestinal failure and long-term use of parenteral nutrition (PN) [56]. It reportedly affects 22%–30% of such children according to clinical diagnoses [57,58]. However, in a recent study, surveillance biopsies in children receiving PN revealed an abnormal histology in 94% and 77% during and after PN, respectively. Of them, 88% and 64% had portal/periportal fibrosis [59]. The causes of IFALD are still being delineated with potential culprits involving plant sterols in PN lipid emulsion, gut dysbiosis, impaired barrier function, and episodes of sepsis. The fibrotic liver milieu in IFALD, even after weaning from PN, has been characterized by α-SMA+HSCs, increased collagen expression, and profibrotic/inflammatory cytokines, such as interleukin (IL)1a, IL1b, epithelial growth factor, adhesion molecule integrin-β6, and matrix metalloprotein (MMP)9 [60-62].

1) Congenital hepatic fibrosis

Congenital hepatic fibrosis is mostly associated with autosomal recessive polycystic kidney disease and a PKHD1 gene mutation that encodes for the transmembrane protein fibrocystin/polyductin, is localized to cilia, and is involved in the development of the renal collecting duct and biliary system [63]. Therefore, the characteristic pathological lesion in the liver is a ductal plate malformation, in which irregular biliary channels persist around the portal tracts [64]. Fibrosis can be severe, forming thick white bands of fibrous tissue across hepatocyte cell islands [65]. Mutation type affecting the PKHD1 gene determines the severity of associated polycystic kidney disease but not congenital hepatic fibrosis [66]. LT outcomes are promising, with good kidney and liver function, even in related living donor LT [67,68].

2) Cystic fibrosis liver disease

Cystic fibrosis (CF) liver disease (CFLD), which develops in 5–10% of patients with CF, is characterized by progressive hepatobiliary fibrosis. In a recent cohort study of 3,328 CF patients, of whom 32.2% and 10% developed CFLD and severe CFLD by 30 years of age, respectively [69]. Lewindon et al. [70] longitudinally followed 41 patients with CF and found that fibrosis at baseline biopsies, but not clinical or imaging abnormalities, heralded the development of clinically significant CFLD/portal hypertension.

However, the patchy nature of disease sampling may be inadequate for quantifying fibrosis. For the noninvasive diagnosis of liver fibrosis, Pereira et al. [71] correlated various serum markers of fibrosis in a cohort of 36 children with CFLD. Lewindon et al. [72] investigated the diagnostic performance of vibration-controlled transient elastography (VCTE) and APRI/Fibrosis-4 (FIB-4) Index for staging fibrosis in 27 and 51 children, respectively. In these studies, VCTE (area under the receiver operating characteristic curve [AUROC], 0.89), APRI (AUROC, 0.83), and prolyl hydroxylase (AUROC, 0.81) were superior to others in determining a meta-analysis of histological data in viral hepatitis (METAVIR) score ≥ F3 and Scheuer stage ≥2 fibrosis [72,73]. Future studies should provide clinical benefits to this patient group owing to the silent progression of fibrosis.

3) Fontan-associated liver disease

Fontan-associated liver disease (FALD) represents a congestive hepatopathy related to elevated central venous pressures that follow the Fontan operation and final-stage surgical palliation for univentricular physiology through complete cavopulmonary shunt creation [74]. According to studies on surveillance biopsies, nearly all biopsied patients develop significant fibrosis during adolescence [75]. Among our cohort of Fontan patients, FALD was detected in 87.5%. Liver stiffness showed a significant positive correlation with post-Fontan duration (P<0.05, r=0.550) and the FIB-4 index (r=0.7, P<0.05) but not with APRI. Splenomegaly (P<0.01, r=0.669), APRI (P<0.01, r=0.767), and FIB-4 index (P<0.01, r=0.922) were all correlated with postFontan duration, implying the importance of universal screening depending on the timing after surgery.

4) Fibrosis in LT

Owing to the improvements in post-LT care, the 10-year patient and graft survival rates in pediatric LT are 83% and 73%, respectively [76]. However, progressive liver fibrosis is among the most common causes of liver allograft failure in the pediatric group, reaching as high as 69%–97% according to different centers [77-81]. Both alloimmune inflammation and biliary outflow obstruction contribute to ongoing fibrosis [5]. Alterations in the gut microbiome and intestinal permeability are potential contributors through pathogen-associated molecular pattern translocation and pattern recognition receptor activation, abnormal bile acid metabolism, and short-chain fatty acid imbalances [82]. According to sequential allograft biopsies in the study by Varma et al. [83], the histological lesion begins with portal inflammation and progresses to periportal fibrosis, followed by extension to adjacent portal fields. Small longitudinal studies have shown that intensification of immunosuppression after histological diagnosis of inflammation and fibrosis leads to a reduction in fibrosis scores [80,84]. To date, studies investigating noninvasive techniques for identifying liver fibrosis in post-LT settings found that VCTE was superior to acoustic radiation force impulse (ARFI) and serum biomarkers (e.g., AST/alanine transaminase [ALT] ratio, APRI, FibroTest [FT], and enhanced liver fibrosis test) [85].

1) Vibration-controlled transient elastography

VCTE is the most commonly used modality for evaluating liver stiffness worldwide, with FibroScan (Echosens, Paris, France) being the most common product [91]. VCTE requires its own probe and relies on creating 50-Hz frequency vibrations to induce shear waves. The velocity at which the generated shear waves travel is directly related to liver stiffness and increases in stiffer tissue [92].

VCTE is the most well-studied modality for estimating liver fibrosis in adult and pediatric settings. In 2018, Hwang et al. [93] performed a meta-analysis of 11 studies involving 723 patients to investigate the correlation between liver fibrosis biopsies and VCTE measurements. The AUROC for detecting stage ≥F2 fibrosis was 0.96, which remained stable across subgroup analyses involving patients aged ≥8 years and study design.

Despite the high accuracy and reproducibility of liver stiffness measurements (LSMs), limitations exist. Studies of healthy children found that LSMs were higher with increasing age and in boys [94,95]. In addition, LSM cutoffs vary by etiology. Studies of heterogeneous patient populations found a cutoff level of 7.5–13 kPa for detecting stage ≥ F3/F4 fibrosis [96-99]. In studies of infants with BA, before KPE, the cutoffs were 9.8–23.2 kPa [100-102]. Therefore, validation studies in large cohorts with subgroup analyses are needed to improve the reliability and accuracy of VCTE in clinical practice.

2) Two-dimensional shear wave elastography and point shear wave elastography

Similar to VCTE, 2-dimensional shear wave elastography (2D-SWE) and point shear wave elastography (p-SWE) use shear waves created by the liver parenchyma that travel with higher velocity in a stiffer medium. In contrast to VCTE, 2D-SWE and p-SWE use ARFI to induce shear waves. Both modalities employ a regular ultrasound probe to take measurements; however, 2D-SWE samples have a larger area than p-SWE samples [103].

A 2016 meta-analysis of 13 articles on 2D-SWE in the diagnosis of liver fibrosis in adults found high sensitivity and specificity of 2D-SWE across all stages of fibrosis [104]. In a meta-analysis of 12 studies of 2D-SWE accuracy in fibrosis in children, Kim et al. [105] computed a summary sensitivity, specificity, and AUROC of 81%, 91%, and 0.93 for detecting stage ≥ F2 fibrosis, respectively. Median cutoff values for detecting significant fibrosis as summarized by Jiang et al. [104] and Kim et al. [105] were 8.0 kPa and 9.4 kPa in adults and children, respectively.

Individual studies have evaluated the diagnostic performance of 2D-SWE in disease groups and patients with potential confounders. Dhyani et al. [106] and Galina et al. [107] compared the accuracy and cutoff values of 2D-SWE across hepatocellular and cholestatic disease groups. They both found higher LSM cutoff values for detecting fibrosis in cholestatic groups. Chen et al. [108] did not report a significant difference between LSM cutoffs in BA and non-BA cholestatic infants, supporting that cholestasis, but not BA alone, impacts LSM. Moreover, Tutar and Alhashmi et al. found that steatosis increases the LSM cutoff and reduces overall performance of 2D-SWE, while in their cohort of 68 patients with NAFLD Garcovich et al. reported AUROC values of 0.92 and 0.97 for detecting stage ≥F1 and ≥F2 fibrosis, respectively [109-111]. Dardanelli et al. [112] identified hemosiderosis as a confounder of 2D-SWE.

Overall, p-SWE has been described as a tool with greater reliability than transient elastography but lower accuracy than 2D-SWE [113]. In their 2016 systematic review, Andersen et al. [114] included 6 studies that evaluated p-SWE versus liver biopsy performance. They found poor distinctive ability for no, mild, and moderate fibrosis and evidence of confounding by necroinflammatory activity. Our search revealed 3 additional publications with AUROC values of 0.83–0.86 for detecting METAVIR scores ≥F3 with cutoffs. Across all studies, the cutoff for detecting F3 fibrosis was 2.0–2.17 m/sec [115-117].

3) Magnetic resonance elastography

Similar to US-based techniques, magnetic resonance elastography (MRE) involves the creation and measurement of shear wave velocity that travels through the liver parenchyma. Propagating shear waves were measured using standard MR phase-contrast imaging sequences with added motion encoding gradients [118]. Performance statistics in previous adult studies were very promising as they found AUROC values of 0.83–0.99 and 0.84–1.0 for detecting stage ≥F2 and ≥F3 fibrosis, respectively [119]. To date, 3 large studies have evaluated the role of MRE in staging liver fibrosis, with AUROC values of 0.53–0.93 [120-122]. All studies were relatively uniform in terms of the patient population (age 13–14.2 years) and disease (mainly hepatocellular pattern). However, the accuracy and reliability of MRE in NAFLD was conflicting between the Trout and Schwimmer et al. studies, with AUROC values of 0.53 and 0.93 for this population, respectively.

1) APRI and FIB-4

The APRI and FIB-4 scores were initially developed to monitor liver fibrosis progression in adults with hepatitis C [123,124]. Both are calculated using basic biochemistry, making them more accessible and cost-effective than radiological modalities.

The APRI was calculated using serum AST and platelet counts. The advancement of fibrosis reduces platelet counts in 2 ways: by reducing thrombopoietin generation by hepatocytes; and hypersplenism. Meanwhile, AST increases as chronic injury drives its release from mitochondria, and fibrosis diminishes clearance [125-127]. This pathophysiological derangement occurs later in the disease course, explaining why APRI has not always been a sensitive tool for detecting mild to moderate fibrosis [33,34]. Finally, APRI requires the use of the standard upper limit of normal for AST, which is 40 IU/L in most pediatric studies. Nevertheless, children often have lower baseline enzyme levels, which introduces potential bias to the score [128].

The FIB-4 index considers age-related differences in enzymes and ALT in addition to the components of APRI [123]. Despite these additional elements, APRI consistently outperformed FIB-4 in detecting fibrosis in both children and adults with liver disease [73,129-135]. One exception was the prediction of portal hypertension in a cohort of patients with CFLD, in which the FIB-4 index (AUROC, 0.91) had a superior AUROC to the APRI (AUROC, 0.71) [73].

2) FibroTest

FT is another composite score developed for the noninvasive evaluation of fibrosis in adults with chronic HCV. The FT combines age, sex, and 5 common serum markers: total bilirubin, a2-macroglobulin, haptoglobin, and gamma-glutamyl transferase [136]. Overall, the validation of FT in the pediatric cohort has been limited to relatively small studies comprising children with various CLD or viral hepatitis and did not reveal consistency [136-140].

3) Pediatric NAFLD Fibrosis Index and Pediatric NAFLD Fibrosis Score

Two indices were developed for noninvasive evaluation of liver fibrosis in children with NAFLD: Pediatric NAFLD Fibrosis Index (PNFI) and Pediatric NAFLD Fibrosis Score (PNFS). PNFI is based on age, waist circumference, and triglyceride levels. In a development cohort of 203 patients, the AUROC for predicting fibrosis was 0.85 [141]. Subsequently, Alkhouri et al. utilized PNFI in combination with an enhanced liver fibrosis test (ELF; including hyaluronic acid, procollagen-3, and tissue inhibitor of metalloproteinases-1) and VCTE. In both studies, ELF and VCTE were superior to PNFI, with AUROC values of 0.618–0.761 for detecting various fibrosis levels [142,143]. Similarly, Yang et al. [134] found that, across multiple diagnostic tests, PNFI had the lowest AUROC value in a cohort of 77 children with NAFLD. In 2014, Alkhouri et al. [131] developed PNFS in a cohort of 242 children with biopsy-proven NAFLD, which included ALT, alkaline phosphatase, platelet count, and gamma-glutamyl transferase levels. The AUROC value for this score was 0.74 for detecting stage ≥F3 fibrosis. However, external validation of this tool is still lacking [144].