What Is Known
- Primary sclerosing cholangitis affects approximately 1.5 per 100,000 children in the United States.
- Concomitant diagnoses with other autoimmune diseases include inflammatory bowel disease in 70% to 80% and autoimmune sclerosing cholangitis (overlap with autoimmune hepatitis) in 25% to 40% of patients with primary sclerosing cholangitis.
- Prognosis in adults with primary sclerosing cholangitis is an estimated 65% 10-year survival.
What Is New
- The 10-year transplant-free survival in this cohort was 89%.
- There were no differences in liver-related outcomes between patients with primary sclerosing cholangitis and autoimmune sclerosing cholangitis.
- Primary sclerosing cholangitis that develops after a diagnosis of inflammatory bowel disease has a milder course in childhood.
Primary sclerosing cholangitis (PSC) affects approximately 1.5 per 100,000 children in the United States, and can lead to severe outcomes including end-stage liver disease and liver transplantation (LT) (1). The presentation of PSC can include biochemical abnormalities and symptoms including jaundice, pruritus, hepatomegaly, and poor growth (2). PSC is associated with other autoimmune diseases, especially inflammatory bowel diseases (IBDs), and can be part of an overlap syndrome with autoimmune hepatitis (AIH). The latter is also known as autoimmune sclerosing cholangitis (ASC) (3).
To date there have been only a few reports of cohorts used to examine the phenotype and natural history of pediatric PSC; these have included relatively small numbers of patients (N < 60) with relatively short duration of follow up (1–6). Prognosis in adults with PSC is considered poor, with an estimated 65% 10-year survival (7). Whether equivalent proportions of children with PSC will develop complications of end-stage liver disease over similar intervals of time is unclear. Longitudinal studies of larger pediatric cohorts are required to clarify long-term outcomes.
At Boston Children's Hospital, a large number of patients with PSC have been seen in consultation during the past 30 years. The present study is an examination of the pediatric PSC phenotype at diagnosis and in longitudinal follow-up, including a description of the clinical presentation, the pattern of liver biochemistry changes over time, the presence of ASC (overlap syndrome), and/or IBD and the proportion of children with liver-related outcomes.
Methods are available online as Supplemental Digital Content 1 (http://links.lww.com/MPG/A755).
Characterization of Primary Sclerosing Cholangitis at Diagnosis
A total of 120 patients were included in this cohort. Following identification of 152 patients, 115 from the hepatology clinic database and 37 from medical records, 32 were excluded: 18 did not have PSC on further evaluation, and 14 had incomplete data. Patients were diagnosed at a median age of 14 years (interquartile range [IQR] 10, 16) and the majority were boys (63%). Median time to PSC diagnosis following presentation with symptoms or abnormal liver biochemistry was 2.5 months (IQR 1.2, 9.8; Table 1). The majority of patients presented with abnormal liver biochemistry (81%, asymptomatic in 64%), whereas a small proportion had symptoms, such as abdominal pain or pruritus, at the time of presentation (27%); few had jaundice or hepatomegaly (6% and 3%, respectively). The median height and weight z scores at diagnosis were not different from those in the normal pediatric population.
At diagnosis, the median GGT (gamma glutamyl transferase) was 221 U/L (IQR 110, 425; upper limit of normal [ULN] 55), and median ALT (alanine aminotransferase) was 125 U/L (72, 194; ULN 30). There were no significant differences in GGT between patients with PSC and ASC (Supplemental Digital Content 2, Table S1, http://links.lww.com/MPG/A756). In those patients concurrently diagnosed with AIH and PSC, ALT and AST were, however, both significantly higher than those with only PSC (ALT—ASC: 216 [138,317], PSC: 123 [60,170], P = 0.02; AST—ASC: 127 [69,254], PSC: 69 [38,111], P = 0.01). Median total bilirubin levels were normal in both the overall cohort and the ASC subgroup (0.4 mg/dL [IQR 0.3, 0.7], 0.7 mg/dL [0.4, 1.4], respectively).
Liver ultrasonography (US), obtained in 83 patients within 6 months of PSC diagnosis, was abnormal in 64 (sensitivity = 77%). Findings included biliary duct abnormalities (N = 44), splenomegaly (N = 20), and gallstones (N = 1). Both the US and the cholangiograms were abnormal in 67% of patients who had both studies performed within 6 months of PSC diagnosis (50/75). Eleven of these 75 patients (15%) had normal US and abnormal cholangiograms; all 11 had intrahepatic abnormalities, whereas 6 of 11 had diffuse large bile duct abnormalities on magnetic resonance cholangiopancreatography or endoscopic retrograde cholangiopancreatography that were not identified on US.
Liver biopsy data were available in 100 patients at the time of diagnosis, 81 of these had cholangiography within 6 months of diagnosis. The diagnosis of PSC was based on findings specific to PSC on both histology and cholangiography in 27% of patients (Table 1). In some cases, histology was consistent with PSC, whereas cholangiography demonstrated irregularities that were abnormal but not specific to PSC, and vice versa. Approximately two thirds of patients (64/97) had large duct involvement at diagnosis (Supplemental Digital Content 2, Table S2, http://links.lww.com/MPG/A756). Isolated extrahepatic large duct involvement was rare (4/64). Isolated intrahepatic large duct involvement was present in 40% of those with macroscopic disease (28/64), and 53% had diffuse, intrahepatic and extrahepatic, large duct disease (34/64). Forty-one percent had advanced fibrosis (METAVIR stage 3 or 4) at the time of diagnosis. No patients with abnormal cholangiograms had normal liver histology, with either nonspecific irregularities, or findings specific to PSC.
Autoimmune Sclerosing Cholangitis (Primary Sclerosing Cholangitis/Autoimmune Hepatitis Overlap Syndrome) at Primary Sclerosing Cholangitis Diagnosis
Of the 120 patients, 31 (26%) had ASC (Table 1); the majority (20/31, 65%) were concurrently diagnosed with PSC/AIH, whereas 7 patients had a diagnosis of AIH more than 6 months after the PSC diagnosis (1.1–8.3 years later), and 4 patients were diagnosed with AIH before the PSC diagnosis (1.3–8.1 years before). Biochemical and radiographic data were not different between patients with PSC and ASC at diagnosis, except for a higher rate of autoimmune marker positivity in the latter (ANA 59% [10/17] vs 16% [6/38], P = 0.003; smooth muscle antibodies 69% [11/16] vs 36% [13/36], P = 0.04). There were only 4 of 18 patients with ASC who were seronegative for both tests. The degree of fibrosis was not significantly different between patients with isolated PSC and those with ASC; marked fibrosis or cirrhosis in 41% (30/74) and 42% (11/26), respectively (P = 0.87). Mean height z scores were lower in patients with isolated PSC compared to those with ASC (−0.19, standard deviation 1.6 vs 0.76, standard deviation 2.1, respectively, P = 0.03).
Small Duct Primary Sclerosing Cholangitis at Primary Sclerosing Cholangitis Diagnosis
Small duct PSC was present in 24% of cases (24/100) at diagnosis. These patients were diagnosed at a younger age compared to the 53 others who had both biopsy and cholangiograms at baseline (10 years [IQR 8, 16] vs 14 years [11, 16], respectively, P = 0.03). Fifty percent (12/24) of patients with only small duct PSC had advanced fibrosis (METAVIR stage 3 or 4) at the time of diagnosis, which was not different when compared to the remaining patients with PSC (38%, P = 0.30). The rate of ASC in patients with small duct PSC compared with the remaining cohort was 38% (9/24) versus. 22% (17/76), respectively (P = 0.14).
Inflammatory Bowel Disease Phenotype at Primary Sclerosing Cholangitis Diagnosis
Investigations for possible IBD were performed in 86% of patients (Table 1). IBD was eventually identified in 81%, with a higher prevalence associated with isolated PSC, compared with ASC (85% vs 68%, P = 0.03). In patients with IBD, there was a higher prevalence of ulcerative colitis (UC) than Crohn disease or IBD-U (70%, 26%, and 4%, respectively). IBD was typically diagnosed either within 3 months, or before, the PSC diagnosis (46% [45/97] and 44% [43/97], respectively); PSC was diagnosed >3 months before the IBD diagnosis in 9% [9/97]. In the 43 patients who were diagnosed with IBD first, the median interval before PSC diagnosis was 298 days (95% confidence intervals [CIs] 235, 634). There was no significant difference in interval before PSC diagnosis in those who were found to have ASC versus PSC (ASC: 291 days [95% CI 102, 1095], PSC: 329 days [95% CI 235, 692], P = 0.71). The degree of liver fibrosis at baseline was not different between those with PSC-IBD and those without IBD; however, there was a lower rate of overlap AIH in patients with PSC-IBD (Table 1).
Pancolitis was identified in the majority of patients with UC/IBD-U (56%, 40/72), with no significant differences between PSC and ASC (P = 0.26). Macroscopic rectal sparing was observed on endoscopy in 12 of 95 patients with any colitis (13%), and in 30% of patients with pancolitis. In those patients diagnosed with IBD within 3 months of the PSC diagnosis, the severity of intestinal disease was mild in patients with UC/IBD-U (median PUCAI 30 [IQR 18, 40], N = 24), and moderate to severe in patients with Crohn disease (median PUCAI 35 [IQR 35, 40], N = 5).
Other autoimmune diseases were identified in 8% of patients (10/120) and included type I diabetes mellitus, thyroid disease, and celiac disease (4%, 3%, and 2%, respectively, Table 1). There were no statistically significant differences in the prevalence of other autoimmune diseases between the PSC and the ASC groups. A family history of PSC or AIH was infrequent (1/98 and 2/103, respectively), compared with a family history of IBD (29/114). There was a high rate of family history of other autoimmune diseases (21/107), such as type 1 diabetes mellitus or thyroid disease, greater in children with ASC (ASC: 32%, PSC 15%, P = 0.05).
Median follow-up from PSC diagnosis was 3.7 years (IQR 1.5, 6.9); 5- and 10-year follow-up was available in 74 and 17 patients, respectively. Longitudinal data were retrospectively collected at intervals of 1 (N = 105), 2 (N = 90), 5 (N = 51), and 10 (N = 17) years after diagnosis. There were no significant baseline differences in sex, presence of ASC overlap, or diagnosis with IBD between the patients who had an LT or had data at 5 years, compared with those with <5 years of follow-up data.
At 1, 2, 5, and 10 years after PSC diagnosis, medication data were available in 90, 80, 48, and 14 patients, respectively. Approximately 80% of patients with PSC were treated with ursodeoxycholic acid (UDCA) at each of these time points. Other antipruritic medications were administered in fewer than 10% before 5 years and 21% (3/14) at 10 years. More patients were taking corticosteroids at 1 and 2 years post-PSC diagnosis (28%) compared with those followed until 5 and 10 years (15%). A greater proportion of patients with ASC were taking corticosteroids than those with PSC (1 year: 57% vs 22%, P = 0.0007; 2 years: 42% vs 22%, P = 0.07; 5 years: 33% vs 8%, P = 0.03); there were no significant differences in corticosteroid use between patients with or without IBD. Between 35% and 46% of patients were taking immunomodulators such as thiopurines or methotrexate for IBD or AIH at each time period. A higher proportion of patients with ASC were taking corticosteroids and immunomodulators compared to patients with isolated PSC.
The most significant changes in liver biochemistry over time were seen in GGT and ALT levels between baseline and 1 year after diagnosis. Median GGT decreased from 221 U/L (IQR 110, 425, N = 87) to 44 U/L (IQR 14, 154; N = 69) and median ALT decreased from 125 U/L (IQR 72, 194; N = 95) to 28 U/L (IQR 16, 71; N = 75); P < 0.0001 for both. There were, however, no significant differences observed between the 1 year and the 2-, 5-, or 10 year time points. Of the 93 patients with liver biochemistry at 1 year after diagnosis, 33% had elevated GGT (girls >35 U/L, boys >55 U/L), with a median GGT of 188 U/L (IQR 78, 317) and 33% had elevated ALT (>40 U/L), with a median ALT of 76 (IQR 47, 137). Within the ASC population, 9 of 21 patients had an ALT >40 U/L at 1 year with a higher median abnormal ALT level of 97 U/L [61, 176], compared with 75 U/L [45, 120] in PSC (P = 0.001). There were no significant differences in mean direct bilirubin, albumin, or international normalized ratio over time.
Of the 110 patients who had cholangiograms at diagnosis, 35 patients had repeat imaging >6 months post-PSC diagnosis. Of these, 7 (20%) never developed any large duct abnormalities. Progression of disease changed from no large duct disease to intrahepatic large duct disease in 3 patients (9%), and to diffuse disease in 7 (20%). Of the 14 patients with diffuse large duct disease at baseline and follow-up imaging, 4 patients demonstrated improvement over time, with no extrahepatic large duct disease seen on follow-up cholangiograms. Findings are summarized in Supplemental Digital Content 2, Table S2 (http://links.lww.com/MPG/A756). Cholangiography was obtained at baseline, 1 to 5 years (median 3.4 y [IQR 1.9, 4.3]), and at >5 years (median 8.7 years [IQR 6.8, 9.5]) after PSC diagnosis in 97, 25, and 16 patients, respectively. The distribution of large duct disease did not change considerably over time. Nine of the 24 patients with only small duct PSC at diagnosis had repeat cholangiograms, and 6 of these demonstrated large duct abnormalities (intrahepatic in 3, intra and extrahepatic in 3), with a median 6 years after diagnosis (IQR 3, 10).
Nineteen patients had more than 1 liver biopsy: 92 at baseline, 19 between 1 and 5 years, and 8 more than 5 years post-PSC diagnosis. The median METAVIR stage was unchanged between baseline and other time points: 2 (IQR 1, 4) at baseline, 2 (IQR 1, 4) at 1 to 5 years post-PSC diagnosis (P = 0.88), and 1 (IQR 1, 3) at >5 years (P = 0.61). Five of the 6 patients (1 had missing data) who went to LT demonstrated progression from F2/F3 to cirrhosis at LT (Table 2).
Over the course of follow-up, 27% of patients developed at least 1 liver-related complication (Table 3). There were 17 patients with cirrhosis detected at any point: 11 at baseline, and 9 during follow-up. Five of the 9 had documented progression to cirrhosis on serial biopsies.
Esophageal varices developed in 14 patients (5 with bleeding), 4 had at least 1 episode of bacterial cholangitis, and 4 developed pruritus.
Nine patients were listed for LT, and 6 patients underwent LT by the time of the present study (Table 2). The 5-year survival with native liver was 89%, when accounting for censoring (Fig. 1). The earliest LT was 2.5 years after the PSC diagnosis. Although there were 5 patients with isolated PSC and only 1 patient with ASC in the LT group, there were no statistically significant differences between the groups (log rank Chi-square 0.53, P = 0.47); this may be due to small absolute numbers. Transplant-free survival for patients with small duct PSC was not significantly different from those with large duct abnormalities by Kaplan-Meier analysis (log rank Chi-square 0.56, P = 0.46). Among patients with GGT data available at 1 year post-PSC diagnosis, an elevated GGT >2 times ULN was detected in 3 of 5 patients (60%) who were transplanted during follow-up and 18 of 87 (21%) of those who were not transplanted (P = 0.08). Of those with medication data available at 1 year post-PSC diagnosis, all 4 patients who underwent transplant and 69 of 85 patients who were not transplanted had received UDCA.
One transplanted patient developed recurrence of PSC in the liver graft. This patient had small duct PSC and IBD at PSC diagnosis, which evolved over the course of 4 years to include large duct. A second patient developed a dominant stricture in the left hepatic duct after LT; however, whether this was due to PSC recurrence versus transplant-related biliary complications remains unclear.
Liver-related outcomes were less frequent in patients with IBD compared with those without IBD (21% of 97 vs 52% of 23, P = 0.002). There appeared to be a lower rate of cirrhosis in patients with IBD versus without IBD, although this was not statistically significant (22% of 93 vs 41% of 22, P = 0.06). There was a lower rate of LT in patients with IBD (2% of 95 vs 18% of 22, P = 0.01).
Patients diagnosed with IBD prior to PSC were compared with all remaining patients in the cohort. There were no differences in age at PSC diagnosis (median 14 years [IQR 10, 17], N = 43, vs 14 years [IQR 10, 16], N = 77, P = 0.78). The degree of fibrosis at baseline was, however, milder, with 32% of patients (12/37) demonstrating marked fibrosis (METAVIR F3) or cirrhosis, compared to 49% in the remaining cohort (27/55, P = 0.03). There was a lower rate of cirrhosis and esophageal varices in patients diagnosed with IBD first, compared to the remaining cohort (cirrhosis: 15% [6/41] vs 31% [23/74], P = 0.05; esophageal varices: 5% [2/41] vs, 20% [15/75], respectively, P = 0.03), but no difference in the rate of bacterial cholangitis. None of the patients diagnosed with IBD first underwent an LT or died during follow-up (Kaplan-Meier transplant-free survival for IBD before PSC vs the remaining cohort: log rank chi-square 4.7, P = 0.03).
The overall mortality rate was 2% (2 patients) in this cohort. The first child had PSC-IBD and was 7 years of age at the time of death. This patient was admitted shortly after his first dose of infliximab, for treatment of clostridium difficile infection with vancomycin, and died suddenly of multiple intracranial arterial strokes. The second patient died at the age of 7 years, 2 years following an LT, complicated by recurrent bouts of acute cellular rejection and post-transplant lymphoproliferative disorder. There were no instances of hepatobiliary malignancy identified in the cohort.
The present study reviews the phenotype and outcomes of PSC in a retrospective longitudinal cohort at a single pediatric institution. To our knowledge, this is the largest reported cohort study, with 120 children with PSC at diagnosis, and data on 51 patients at 5 years of follow-up. We demonstrated that PSC is clinically silent in the majority: 64% presented with abnormal liver biochemistry and no other symptoms. Small duct PSC was diagnosed in 24% of cases, and demonstrated similar outcomes compared to patients with large duct PSC. Within the subset of patients who had repeat biopsies or imaging in follow-up, there were no significant changes in degree of liver fibrosis or distribution of large duct abnormalities over time. PSC-IBD demonstrated a milder liver disease phenotype, and the subset who were diagnosed with IBD first had a lower frequency of cirrhosis and no need for LT compared to the overall cohort. Overall, the 5-year survival with native liver was higher in this cohort (89%) compared to that reported in adults (7).
Liver biochemical tests improved by 1 year post-PSC diagnosis in this cohort. This was likely related to the use of UDCA, as the majority of patients (82%) were using this medication. UDCA has been shown to cause biochemical improvement in patients with PSC; however, this has not correlated with better long-term outcomes in adults (8). UDCA did not appear to be specifically protective against liver transplantation, and the number of patients transplanted was too small to comment on whether UDCA confers a higher risk of liver transplantation in this population. The GGT and ALT levels did not increase over time in our overall cohort, or in the children who required an LT. These biochemical markers are likely of little use in the assessment of progressive liver disease in pediatric PSC. Furthermore, ALT remained above the ULN in 33% of children with PSC and 43% of ASC. This has implications in the ASC population in whom immunosuppressants are used to treat AIH. Patients in this cohort with elevated liver biochemistry at 1 year post-PSC diagnosis did not routinely have repeat histology or autoimmune marker assessment. It may be prudent to consider re-evaluation at the 1-year mark in those with elevated liver biochemistry to distinguish uncontrolled AIH from pure PSC disease to tailor immunosuppression strategies to those with uncontrolled AIH. Achievement of normal liver biochemistry in PSC may be unattainable, the pursuit of which may lead to inappropriate immunosuppression and adverse effects.
Liver US was 77% sensitive for any abnormality in 83 children with PSC in this cohort; a minority with large duct disease on cholangiography had a normal US, demonstrating lack of utility of this modality in ruling out pediatric PSC. Furthermore, small duct PSC was present in 24%. On the contrary, all patients with abnormal cholangiography had abnormal liver histology. In pediatric patients with significantly elevated GGT, a liver biopsy is a useful investigation in the evaluation of possible PSC, whether or not IBD is present.
The rates of stent placement and balloon dilation within the first 6 months of diagnosis were presented in terms of the total number of endoscopic retrograde cholangiopancreatography procedures performed (N = 36). When compared to the total number of children in this PSC cohort (N = 120), the rate of stenting and balloon dilation were, however, 3% and 7%, respectively. A review of previously published pediatric PSC cohorts (N = 29–52) revealed stenting and balloon dilation rates of 4% to 8% and 2% to 12%, respectively, although the timing for these procedures was not reported (1,2,5). The rates of interventional cholangiography in this group were similar to those reported in other cohorts.
Children with PSC-IBD of the UC or IBD-U phenotype demonstrated mild severity of intestinal disease, pancolitis in 56%, and macroscopic rectal sparing in 13%. This is different from what is observed in adults with PSC-IBD where pancolitis has been reported in 87% and rectal sparing in 52% (9). The intestinal phenotype of PSC-IBD in children may be less specific than that observed in adults.
Healthy children are not screened for liver diseases with routine biochemical tests at well-child visits, per recommendations by the American Academy of Pediatrics (10). As many patients with PSC present with asymptomatic elevations in liver enzymes, it is possible that some of these children had PSC for an extended period of time before diagnosis. Alternatively, children with IBD are frequently surveilled with liver biochemistry on a routine basis. This may be the reason that the PSC subset who were diagnosed with IBD first were found to have milder liver disease at diagnosis. In addition, they had a lower rate of cirrhosis, and no need for LT during follow-up. This might be an opportunity for pharmaceutical research in PSC management, since by targeting early-onset PSC, the development of treatments that can prevent progression to end-stage liver disease might be feasible.
Limitations of the present study include its retrospective design, with no protocol liver biopsies or imaging. Although data at diagnosis was usually complete, the numbers at follow-up decreased, due to transition to adult care or other facilities. Despite this, the longitudinal analysis was robust as there was a large amount of data available at 5 years of follow-up, and some information at even later time points. Another limitation to the present study was the inclusion of patients from regional clinicians who were seen only once at this institution for consultation. This provided a large volume of descriptive information at PSC diagnosis, but could have led to an under-representation of liver-related outcomes. Because this hospital is the major referral center for LT in the region, we, however, suspect that patients followed by local gastroenterologists would have been referred back with the development of end-stage liver disease and need for transplantation. Thus, we hypothesize that the rate of transplant-free survival is likely fairly accurate. Furthermore, there were no significant differences in phenotype (sex, presence of ASC-IBD) between those patients with available data on outcomes compared to those with missing data at 5 years.
Pediatric-onset PSC has a milder phenotype compared to adult-onset PSC. As PSC can be clinically silent for many years before diagnosis, it is possible that some instances of diagnosis in young adults represent late diagnoses of pediatric PSC. In this cohort of children with PSC, 10% had cirrhosis at the time of PSC diagnosis, compared to 30% in adult studies (11) and some developed cirrhosis during the period of observation. Children have a higher transplant-free survival at 5 and 10 years post-PSC diagnosis, compared to adults. These findings are important in the counseling of children and family members diagnosed with PSC.