Primary sclerosing cholangitis (PSC) is a chronic, progressive, cholestatic liver disease characterized by inflammation and obliterative fibrosis of the intrahepatic and/or extrahepatic biliary tree, resulting in multifocal strictures and dilatation. It is a rare disease, with an incidence and prevalence of 0.1–0.2 and 1.5 per 100,000 children, respectively, which is substantially lower than in adults [11–13]. Pediatric PSC typically presents early in the second decade of life and has a modest male predominance, as in adults [12, 14–16]. The link between PSC and IBD has been known for greater than five decades [17]. As many as 60–80% of adults with PSC in North America and Northern Europe have IBD, primarily ulcerative colitis (UC) [18, 19]. The prevalence of IBD in children with PSC is also very high, >50% in most series and up to 97% in a recent population study [12, 14–16, 20]. Conversely, only a minority of children with colitis, <10% in most series, have or develop concurrent PSC [7, 12, 21, 22]. However, these reports may underestimate the true prevalence of PSC in IBD as neither adult nor pediatric IBD patients are systematically investigated with liver biopsy and cholangiography to screen for liver disease. Most patients are found to have PSC within a year of their IBD diagnosis [12], but the two can occur years apart, and PSC can manifest first, in which case a full colonoscopy is recommended to screen for IBD [23].

The pathogenesis of PSC remains incompletely understood. Genomewide association studies have identified a number of HLA and non-HLA risk loci [24, 25], some of which are shared with IBD, and a hallmark paper in 2004 reported an accumulation of gut-homing CCR9-positive T cells in explanted human livers of patients with PSC [26], findings that point to both a genetic and immunological basis for PSC. In addition, there is growing evidence for the role of the “gut-liver” axis in the pathogenesis of PSC. Several animal models and human tissue-based translational studies support that enteric microbial molecules/dysbiosis can lead to PSC-like hepatobiliary inflammation [27].

The intestinal inflammation in individuals with PSC and colitis may represent a distinct IBD phenotype, termed PSC-IBD. This has been well characterized in adults as extensive colonic involvement, often worse on the right, and relatively frequent “backwash ileitis,” rectal sparing, and pouchitis post colectomy [28]. Crohn disease (CD) is uncommon in the setting of PSC, but, when it does occur, it too tends to have an extensive colonic distribution; isolated small bowel, perianal, and fistulizing disease are uncommon [29]. Despite the extensive nature of the colonic inflammation, PSC-IBD tends to have a relatively mild clinical course with a paucity of overt clinical symptoms [30, 31]. There are significantly less data pertaining to the phenotype of pediatric PSC-IBD, but findings analogous to those in adults have been reported in two small studies [32, 33]. In addition, a recent study specifically aimed at investigating the phenotype of pediatric PSC-IBD compared 37 children with IBD and PSC or ASC to 137 non-PSC matched IBD controls. In keeping with the above, the authors found a higher proportion of pancolitis in the PSC-IBD group, although this was only marginally statistically significant. In contrast to some of the adult evidence, both groups were similar in terms of the proportion of patients with rectal sparing (defined histologically) and disease activity, as reflected by physician’s global assessment, Mayo endoscopic scores, admission rates, and colonic surgery rates [34].

The interplay between IBD and PSC remains to be elucidated. Adults with severe PSC requiring liver transplant (LT) have been found to have milder UC than patients with less severe liver disease, suggesting that PSC may have a “protective” effect on colonic disease [35]. Furthermore, while it has long been maintained that PSC and IBD progress independently, as supported by older studies indicating that the natural history of PSC is unaffected by colectomy [36], more recent findings suggest that colectomy may reduce the risk of PSC recurrence post LT [37]. The interaction between PSC and IBD, including the effect of ongoing colonic inflammation on PSC progression, if any, requires further clarification.

The diagnosis of PSC in a child is based on a compatible clinical presentation and biochemistry, with characteristic changes on cholangiography and/or liver biopsy, after excluding secondary causes of sclerosing cholangitis. The most common presenting symptoms and signs are hepatomegaly and abdominal pain, followed by diarrhea, splenomegaly, fatigue, pruritus, weight loss, impaired growth, and jaundice [15]. The presenting features may also, uncommonly, be those of advanced liver disease, such as gastrointestinal bleeding and cholangitis, or those of associated colitis, especially bloody diarrhea. About 20% of children with PSC are asymptomatic at presentation and come to medical attention solely due to deranged liver biochemistry. Transaminases are often modestly elevated, with a predominantly cholestatic pattern [15]. GGT is more reliable in children as ALP elevations may reflect bone growth. The odds of PSC are 660-fold greater in children with ALT and GGT elevations >50 U/L within 3 months of their IBD diagnosis compared to children whose values remain <50 U/L [9]. INR, albumin, and conjugated bilirubin, which reflect synthetic function, are generally normal at presentation. Elevated conjugated bilirubin may signal a stricture, cholangitis, or a mass, and warrants further work-up. Serum immunoglobulin G (IgG) levels may be elevated, and a variety of autoantibodies may be present, the most common of which is antineutrophil cytoplasmic antibody (ANCA), usually with an atypical perinuclear (“p”) pattern, which is found in up to 80% of patients. None of these are specific to PSC, however [12, 23]. Serum IgG4 should be measured at least once in children with PSC. An elevated IgG4 may denote IgG4-associated cholangitis (IAC), which has important implications, given its favorable response to corticosteroids [38]. Ultrasound is a reasonable initial imaging modality; it may reveal bile duct wall thickening, focal bile duct dilatations, and/or gallbladder changes, including wall thickening, enlargement, cholecystitis, and mass lesions. It is also useful for ruling out alternate etiologies. However, none of these findings are diagnostic, and ultrasound may be normal in the setting of PSC [23]. Cholangiography, preferably by magnetic resonance cholangiopancreatography (MRCP), which has supplanted endoscopic retrograde cholangiopancreatography (ERCP) as the first-line diagnostic imaging modality due its less invasive nature and lower cost, is a vital component of the PSC diagnostic work-up [39]. Characteristic cholangiographic findings include multifocal, short strictures alternating with normal or dilated segments, producing a “beaded” appearance (Fig. 11.1) [23]. The gallbladder, cystic duct, and pancreatic duct may also be abnormal [40]. Contrary to adult guidelines, a liver biopsy is almost always performed in a child with suspected PSC. Histopathological assessment is necessary to distinguish PSC from autoimmune sclerosing cholangitis (ASC), which occurs frequently in children and mandates different treatment. A liver biopsy is also useful to diagnose small-duct PSC, a label applied to cases with compatible histological changes but without cholangiographic abnormalities, and to stage the degree of fibrosis. Periductular concentric fibrosis, or “onion-skinning” (Fig. 11.2), is pathognomonic for PSC, but not always observed. Other, nonspecific findings may include ductular proliferation or periductular inflammation, with variable types of portal inflammation and fibrosis. The diagnostic work-up for suspected PSC in children is illustrated in Fig. 11.3.

PSC is one of the most important sources of morbidity and mortality in IBD, but few studies have examined its natural history in children. Based on limited data, the probability of developing complicated liver disease, defined as clinical portal hypertension, obstructive cholangitis, cholangiocarcinoma, liver transplant, or death, over 5 years in children with PSC, is 37% [12]. About 20% of children with PSC ultimately require a liver transplant (LT), a figure that has remained fairly constant over the past 20 years. The median time from diagnosis to LT is 7–12 years [ 14–16, 41]. Survival is significantly shorter in children with PSC compared to age-matched and gender-matched American children. Lower platelet count, splenomegaly, and older age are associated with shorter survival [14]. Adults with UC and PSC have an almost five times greater risk of colorectal neoplasia compared to adults with UC alone [42]. Surveillance colonoscopies every 1–2 years from the time of diagnosis are recommended in adults [23]. No equivalent pediatric guidelines exist, but it seems reasonable for similar screening practices to be applied to older children and teenagers. There is a markedly increased risk of cholangiocarcinoma in adults with PSC [43], but this malignancy is exceedingly rare in children. Nevertheless, a handful of cases has been reported in older teenagers [12]. While adult guidelines suggest consideration be given to screening for cholangiocarcinoma with regular cross-sectional imaging and CA 19-9, this is not routinely recommended in children [23, 39]. However, based on clinical experience and expert opinion, the authors suggest an ultrasound yearly, an MRI every 2 years, and CA 19-9 levels yearly in children with PSC to screen for cholangiocarcinoma.

Small-duct PSC may have a more favorable prognosis than classic PSC. It has been associated with a longer transplant-free survival in adults, and there have been no reports of cholangiocarcinoma occurring with small-duct PSC. However, small-duct PSC can progress to classic PSC with cholangiographic abnormalities over time, and it can recur post transplant [44]. It is unclear whether small-duct PSC represents an early stage of classic PSC or a distinct entity.

Data pertaining to the medical management of PSC in children are scarce, and current practices largely derive from adult studies. No medical therapy currently exists to reverse or halt the progression of PSC liver disease. As such, treatment is mainly supportive. Although numerous aspects of PSC invoke an autoimmune basis for the disease, thus far, no single immunosuppressive or immune-modulating agent has been found to be efficacious [45].

Ursodeoxycholic acid (UDCA) is widely used in adults and children with cholestatic liver disease, including PSC. Although biochemical improvement has been demonstrated in children, a beneficial effect on the natural history of PSC, as reflected by a decrease in mortality and/or LT rates, has never been shown [14, 15, 46]. Similarly, adult studies have documented improvements in biochemistry, but not in hard outcomes [47]. Furthermore, the use of high-dose UDCA >28 mg/kg has been associated with a twofold increased risk of death/transplant and a fourfold increased risk of colorectal cancer in adults [48]. There is no consensus regarding the use of UDCA in adults with PSC, with one expert group advising against its use entirely [23] and another merely recommending against the use of high doses [39]. In light of this, it appears prudent to avoid high-dose UDCA in children with PSC, but continued use of low-to-moderate doses, not exceeding 20 mg/kg/day, is reasonable.

Limited anecdotal evidence supports the use of oral vancomycin for treating pediatric PSC [49–51]. Oral vancomycin’s therapeutic effect may occur through immunomodulation, by increasing transforming growth factor-β (TGF-β) and peripheral levels of regulatory T cells [52]. Data from prospective pediatric trials are pending. Metronidazole and minocycline, but not rifaximin, have also been associated with improved liver biochemistry in adults with PSC [53–55]. At the current time, the use of oral antibiotics for pediatric PSC remains experimental, as a benefit beyond biochemical has yet to be demonstrated.

Dominant strictures are less common in children than adults, but should, when identified in association with symptoms or signs such as cholangitis, jaundice, pruritus, right upper quadrant pain, or worsening biochemistry, be managed with ERCP and balloon dilatation, often with sphincterotomy, with or without stent placement [23]. This may prolong symptom-free intervals prior to LT [56]. Although cholangiocarcinoma is rare in pediatrics, brush cytology in the setting of a dominant stricture remains important. ERCP should be performed by a physician who is adequately trained in and experienced with the procedure, which often requires collaboration with an adult gastroenterologist.

Liver transplant remains the only definitive treatment for PSC and should be considered for children with decompensated cirrhosis, recurrent or chronic cholangitis refractory to ERCP, hilar cholangiocarcinoma, and intractable pruritus [23, 57]. PSC accounts for 2.6% of pediatric transplants [58]. The mean age at transplant is 12.6 years. Patient and graft survival after LT for PSC is comparable to that for non-PSC pediatric indications, with 1-year and 5-year patient and graft survival rates of 99% and 97%, and 93% and 76%, respectively. However, a diagnosis of IBD prior to LT is associated with an increased risk of death post LT. Intrahepatic biliary strictures and cholangitis are more common in the first 6 months post LT in children with PSC compared to other liver diseases [59]. Furthermore, PSC recurs in about 10% of children post LT [14, 59, 60]. A diagnosis of IBD and younger age have been linked with an increased risk of PSC recurrence [59, 60]. As mentioned above, colectomy prior to or during LT may decrease the risk of PSC recurrence [37].

IgG4-associated cholangitis (IAC) is a rare inflammatory disorder of the biliary tree, characterized by elevated serum IgG4 levels and infiltration of IgG4+ plasma cells in the bile duct walls, causing thickening and stenoses. IAC is often associated with type 1 autoimmune pancreatitis (AIP), the pancreatic manifestation of IgG4-related disease (IgG4-RD), a systemic multiorgan disorder only defined during the last decade [76]. The typical IAC/IgG4-RD patient profile is that of an elderly man with obstructive jaundice, weight loss, and abdominal discomfort. However, IAC occurring in association with UC has been reported, including in children [77]. The clinical and cholangiographic presentation of IAC is often indistinguishable from that of PSC. Furthermore, 9–36% of patients with PSC have elevated serum IgG4 levels (although usually lower than in IAC) [78, 79], and IgG4+ plasma cells have been documented on liver biopsy in PSC patients [80], further blurring the relationship between the two. However, PSC and IAC appear to be distinct entities, as evidenced by their very different response to corticosteroids; in contrast to PSC, IAC typically shows excellent response to immunosuppressive treatment, including resolution of strictures. However, relapse is common after tapering immunosuppression; long-term low-dose therapy with corticosteroids/azathioprine is often needed, analogous to the management of autoimmune hepatitis [81]. Diagnostic criteria have been proposed for IAC; these combine biochemical, radiographic, and histopathological characteristics with multiorgan involvement of IgG4-RD and responsiveness to immunosuppressive treatment [82].

Figure 11.4 graphically depicts the relationship between autoimmune liver diseases and IBD.

A recent systematic review and meta-analysis examining 32 randomized controlled trials, including a total of 13,177 adults primarily with rheumatological indications for treatment, demonstrated an increased risk of liver enzyme abnormalities in patients treated with methotrexate compared to a comparator agent, but no difference in the risk of liver failure, cirrhosis, or death [83]. The results of two adult IBD studies, in which fairly large numbers of liver biopsies were performed, also found very low rates of hepatic fibrosis in patients receiving methotrexate [84, 85], indicating that hepatic fibrosis is not as commonly observed in methotrexate users as suggested by older studies.

Pediatric IBD studies have found varying rates of biochemical liver abnormalities in children treated with methotrexate, ranging from 10% in a recent systematic review to 39% in a multicenter retrospective comparison of oral and subcutaneous methotrexate. Most resolved spontaneously or with dosage adjustment; medication discontinuation was required in only a minority (<5%) [86–88]. These studies are limited, however, by their retrospective nature, the inability to correlate biochemistry with histopathology, and the inability to definitively ascribe the biochemical abnormalities to methotrexate given the absence of documented normal laboratories prior to medication initiation in most cases. Conflicting data exist regarding whether higher methotrexate doses and parenteral versus oral administration are associated with a greater risk of hepatotoxicity [84, 87, 89]. The risk of hepatotoxicity may be higher in the immediate period after starting methotrexate [90]. Importantly, abnormal liver biochemistry does not reliably identify methotrexate-associated fibrosis.

Based on the available evidence, when initiating methotrexate in children with IBD, the authors recommend obtaining liver biochemistry at baseline, weekly for the first month and every 2–3 months thereafter. In cases of persistent moderate enzyme elevations (up to 2–3× ULN), the dose of methotrexate can be adjusted, preferably in consultation with a pediatric hepatologist, whereas, when faced with more marked elevations (>5× ULN), methotrexate should be entirely held, at least temporarily. A liver biopsy should be performed in cases in which liver enzymes remain abnormal despite medication cessation, or when methotrexate discontinuation would be deleterious to IBD management. The use of methotrexate in patients with underlying liver disease, such as PSC, should generally be avoided, if possible.

Azathioprine (AZA) is a prodrug for 6-mercaptopurine (6-MP), which is, in turn, converted to 6-thioguanine (6-TG), the final effector metabolite. The enzyme thiopurine methyltransferase (TPMT) catalyzes the formation of 6-methylmercaptopurine (6-MMP) and 6-methylmercaptopurine ribonucleotides (6-MMPR) [91]. A systematic review, including 34 mostly adult IBD studies, found a mean overall prevalence of AZA/6-MP-induced “liver disorder” of 3.4% and a mean annual rate of abnormal liver tests (up to 2× ULN) per patient-year of 1.4%, suggesting that thiopurine-associated hepatotoxicity is relatively uncommon. However, most studies did not provide definitions for “liver disorder” and were retrospective in design [92]. Two large pediatric studies examining the use of thiopurines in IBD also found fairly low rates of hepatotoxicity, namely, 4.6% and <3%, respectively [93, 94].

Thiopurine-induced hepatotoxicity can be grouped into three syndromes: (1) hypersensitivity reactions; (2) idiosyncratic cholestatic reactions; and (3) presumed endothelial cell injury. Hypersensitivity reactions usually have their onset within 2–3 weeks. Non-allergic cholestatic injuries are characterized by increased serum bilirubin and ALP, with or without moderate aminotransferase elevations, and typically occur within 2–5 months of therapy initiation. Variable parenchymal cell necrosis is typically seen on liver biopsy. Jaundice regression is not universal upon medication cessation [92]. Nodular regenerative hyperplasia (NRH), peliosis hepatis, sinusoidal dilatation, and sinusoidal obstruction syndrome (SOS, or veno-occlusive disease) fall into the latter category and are felt to be dose-dependent. The inciting injury in this group of vascular pathology is at the level of the endothelial cells lining the sinusoids and terminal hepatic venules and tends to occur between 3 months and 3 years of treatment [95]. More specifically, NRH is thought to result from areas of hepatocyte hypoperfusion and atrophy alternating with adaptive hepatocyte hyperplasia. IBD patients treated with AZA have a cumulative incidence of NRH of approximately 0.6 and 1.3% at 5 and 10 years, respectively [96]. Patients with NRH may be asymptomatic with normal or only mild elevations in liver function tests or isolated thrombocytopenia, or may present with clinically evident portal hypertension (PHT). NRH can be detected on liver biopsy, which demonstrates diffuse transformation of normal hepatic parenchyma into small, regenerative nodules with little or no fibrosis [97], and on MRI, which shows multiple fine, nonenhancing nodules [98]. The course is usually indolent, but, rarely, NRH may progress to end-stage liver disease requiring LT [99]. NRH has also been postulated to be a preneoplastic condition, which may predispose some individuals to developing hepatocellular carcinoma [100]. Thiopurine cessation in patients with NRH is generally followed by biochemical normalization, but patients with PHT have a variable course, with resolution of PHT in some, but persistence in others. Peliosis hepatis results in multiple cystic blood-filled spaces in the liver, spleen, lymph nodes, and other organs, which can lead to hepatic hematomas and, rarely, hepatic rupture [101]. SOS typically presents with a Budd-Chiari like picture, with the triad of rapid-onset ascites, painful hepatomegaly, and jaundice.

A reasonable monitoring strategy when initiating thiopurine therapy might include liver biochemistry at baseline, weekly for the first month, biweekly for the second and third months, and monthly thereafter. 6-MMP levels >5700 pmol/8 × 10⁸ red blood cells have been linked with liver toxicity in children [102], but this finding has not been consistent across all studies [103]. If available, metabolite levels may be used to complement liver enzyme monitoring, and TPMT genotype or activity may be determined prior to initiating therapy, but this remains controversial. Mild liver enzyme abnormalities in children on thiopurine therapy may be observed with repeat blood work, but the authors suggest that the dose of thiopurine be reduced by about 50% in patients with more marked derangements. If this does not result in biochemical normalization after several weeks to months, therapy should be withdrawn entirely. Immediate thiopurine discontinuation should be the approach in any patient with clinically overt jaundice. Liver biopsy should be considered if liver tests fail to normalize after medication withdrawal or if there is any suggestion of PHT, even in patients with normal laboratory parameters.

Based on postmarketing surveillance, the Food and Drug Administration (FDA) has issued warnings about the potential risk of serious liver injury with the use of anti-TNFα antibodies [104]. TNFα plays an important role in many aspects of immune response regulation. The association between anti-TNFα use and the development of autoantibodies is well known, although the pathological role of these antibodies remains unclear [105]. Anti-TNFα related hepatotoxicity does not appear to be dose-dependent, but instead is thought to occur in genetically susceptible individuals who generate an idiosyncratic immune response after inhibition of the TNFα pathway. The release and presentation of hepatic autoantigens by immune cells may be involved [106].

Infliximab (IFX) and adalimumab (ADA) have been implicated in drug-induced liver injury (DILI) in both rheumatology and IBD populations. The median latency period is 13–18 weeks, but is hugely variable; DILI may have its onset after a single infusion/injection, but 20% occur more than 6 months into therapy [107, 108]. DILI seems to occur more frequently with IFX than ADA; the rate of DILI has been found to be 1/120 IFX-treated patients compared to 1/270 ADA-treated patients [109]. This is in keeping with the findings of a large retrospective review of adult IBD patients, in which IFX accounted for a disproportionate fraction of the 2.7% of patients who developed significant liver enzyme elevations felt to be secondary to anti-TNFα therapy [108]. The most common presentation is an autoimmune phenotype with primarily hepatocellular injury, high rates of autoantibody (especially ANA) positivity, and histological findings compatible with autoimmune hepatitis. However, mixed nonautoimmune and predominantly cholestatic patterns also occur. Cases with autoimmune features may have a longer latency and higher peak ALT [107]. Autoantibody positivity prior to anti-TNFα initiation does not appear to predict the risk of DILI [109]. Cases of DILI with AIH features should be managed with anti-TNFα discontinuation, in which case the prognosis is favorable. Some patients benefit from treatment with corticosteroids [107]. Anti-TNFα associated DILI does not seem to be a class effect, and switching to a different anti-TNFα, with close observation, appears safe. Milder cases of hepatotoxicity without overt autoimmune features often resolve spontaneously without anti-TNFα discontinuation [108]. No data currently exist regarding anti-TNFα associated liver injury in pediatric IBD.