Surveillance of primary sclerosing cholangitis – a comparison of scheduled or on-demand ERCP with annual MRI surveillance: a multicenter study

Abstract

Background

Primary sclerosing cholangitis (PSC) is associated with a high risk of hepatobiliary malignancy, especially cholangiocarcinoma (CCA). There are no good tumor markers to screen for CCA, and current recommendations for PSC monitoring are mainly based on expert opinions. The optimal strategy to assess disease progression and screen for CCA – the main cause of death of PSC patients – remains unclear. We aimed to compare three different surveillance strategies and their effect on patient outcomes.

Methods

Data from three distinct PSC cohorts with different surveillance strategies – scheduled endoscopic retrograde cholangiopancreatography (ERCP), annual magnetic resonance imaging/cholangiopancreatography (MRI/MRCP) surveillance, and on-demand ERCP according to ESGE/EASL guidelines – was collected. Patients with PSC diagnosed in 1990 or later were included and the last day of follow-up was 31 December 2023. The composite end point consisted of hepatobiliary malignancy, liver transplantation, or liver-related death.

Results

1629 PSC patients were included, with a median follow-up of 8–11 years. The cumulative incidence of the composite end point was lowest in the group undergoing scheduled ERCP (14.1%, 95%CI 12.0%–16.4%) and highest in the on-demand ERCP cohort (35.0%, 95%CI 28.4%–42.0%). Although the cumulative incidence of CCA was lower in the scheduled ERCP group than in the other groups, it did not differ statistically significantly from the MRI/MRCP surveillance group. No differences were seen in liver-related deaths between the surveillance strategies.

Conclusions

In this study comparing scheduled ERCP, annual MRI/MRCP surveillance, and on-demand ERCP, the strategy based on scheduled ERCP using individual risk stratification is associated with better overall prognosis and outcome.

Introduction

Primary sclerosing cholangitis (PSC) is a chronic inflammatory disease of the biliary epithelium that leads to strictures of the intra- and extrahepatic bile ducts and eventually to cholestasis and secondary biliary cirrhosis [1]. The disease course is variable and can be complicated by recurrent bacterial cholangitis and the development of cirrhosis with further complications. The chronic inflammation is associated with increased proliferation of biliary epithelial cells, and a markedly increased risk of biliary dysplasia and the development of cholangiocarcinoma (CCA), with the relative risk ranging from 161- to 973-fold [2] [3]. The cumulative risk of CCA after 10, 20, and 30 years of PSC is 6%, 14%, and 20%, respectively [4]. The prognosis of CCA is dismal, with a 5-year survival of under 10% and the time from diagnosis to death in PSC patients with CCA reported to be 1.6 years [5]. The overall mortality in PSC patients is elevated, with a standard mortality ratio (SMR) of 2.8–4.2 [4] [5], with the leading cause of death being CCA [4]. In addition, the risk for hepatocellular carcinoma (HCC) is increased, with a lifetime prevalence of 0.3%–2.8% [6], and the cumulative 10-year risk of developing gallbladder carcinoma is around 2% [7]. The median survival of PSC patients from diagnosis until liver transplantation or PSC-related death is 21.3–21.9 years [4] [5].

The disease course of PSC is variable and difficult to predict [8]. There are challenges in terms of markers of progression of bile duct change, to identify those patients who will develop cirrhosis or a hepatobiliary malignancy. The recent European Association for the Study of the Liver (EASL) Practice Guidelines on PSC recommended risk assessment at the time of diagnosis and sequentially based on phenotypic factors and non-invasive tests, including: standard biochemical tests; magnetic resonance imaging (MRI) with magnetic resonance cholangiopancreatography (MRCP); and liver elastography or serum fibrosis tests every 6 or 12 months, depending on risk stratification [9]. For monitoring hepatobiliary malignancy, it is suggested to perform liver ultrasound and/or abdominal MRI/MRC annually. ERCP should be avoided for the diagnosis of PSC and is not recommended for surveillance of disease progression or cancer surveillance. ERCP should be considered if the patient's symptoms worsen, cholestatic liver enzymes rapidly increase, or if there is a new dominant stricture or progression of existing dominant stricture shown on MRI/MRCP.

However, in a recent study, annually performed MRI/MRCP followed by ERCP was shown to be ineffective in detecting CCA early enough to benefit long-term survival [10], but PSC patients undergoing scheduled surveillance with imaging and/or ERCP have been shown to have lower overall mortality rates than those not undergoing scheduled imaging (8% vs. 23%) [11]. The optimal strategy to assess disease progression and monitor for the development of hepatobiliary malignancy, especially CCA, the main cause of death of PSC patients, is still unclear.

The objective of this study was to evaluate the outcomes of PSC patients subjected to three different follow-up strategies, focusing on a composite end point: liver transplantation, hepatobiliary malignancy, and liver-related death. We compared scheduled endoscopic surveillance with brush cytology, the EASL recommendations for on-demand ERCP, and annual MRI/MRCP surveillance followed by ERCP as indicated. The study encompassed three patient cohorts from Finland, the Netherlands, and Sweden.

Methods

Individual patient cohorts

Scheduled ERCP

This prospectively collected cohort consisted of 1048 PSC patients retrieved from the Helsinki University Hospital (HUH) PSC registry in Finland. The registry was originally founded in 2009 and comprises PSC patients referred for ERCP for: (i) documentation of the diagnosis after MRCP, and for individual risk stratification of PSC progression; (ii) follow-up of bile duct disease progression; and (iii) dysplasia surveillance. The patients were referred from the special responsibility area of the Helsinki and Uusimaa hospital district, comprising 2.2 million people (39% of the Finnish population), as well as other central and university hospitals around the country.

Images were evaluated using the Helsinki score (modified Amsterdam score) [12]. Dominant stricture was defined as described earlier [13]. Dilation was performed when a dominant stricture was diagnosed, or if the cytology brush could not be passed through a stenosis.

At HUH, a structured and systematic approach for diagnostics, follow-up, and dysplasia surveillance of PSC patients was implemented in 2009, based on ERCP with brush cytology [14] [15] (Figs. 1s and 2s, see online-only Supplementary material). The PSC diagnosis was ascertained by ERCP. The follow-up and surveillance examinations were scheduled individually based on the severity of the endoscopic findings (Helsinki score), need for possible dilation, degree of biliary inflammation assessed by brush cytology, biliary calprotectin, and IL8, in addition to the presence or absence of biliary dysplasia. The time interval between successive ERCP examinations can vary from 3 months to 5 years. MRCP is done before the next ERCP. The mean number of ERCP examinations per patient was four (range 1–17).

Clinical data and blood samples were collected on the day before or at the time of ERCP. Plasma and serum clinical chemistry parameters were provided by the Helsinki University Central Laboratory (HUSLab). The last day of follow-up was 31 December 2023.

Annual MRI/MRCP surveillance

This prospective patient cohort, consisting of 448 PSC patients, was collected from 10 Swedish hospitals. Patients were enrolled between 1 November 2011 and 1 April 2016 in a 5-year surveillance program, which included yearly contrast-enhanced MRI/MRCP, clinical examinations, liver function tests, and analysis of the tumor marker carbohydrate antigen 19–9 (CA19–9) [10].

Inclusion criteria were a diagnosis of PSC based on cholestatic liver biochemistry, with typical cholangiographic features on MRCP and/or a liver biopsy, age 18 years or older, and an MRI/MRCP at baseline. Exclusion criteria were expected listing for liver transplantation within 1 year after inclusion, previous liver transplantation, and/or the presence of a hepatobiliary malignancy. Diagnosis of cirrhosis at the time of inclusion was based on previous histology or clinical/radiological features of portal hypertension. At the time of inclusion, data on age, sex, co-morbidities, previous medical history, inflammatory bowel disease (IBD), medications, liver function tests, CA19–9, and data from a baseline MRI/MRCP were collected. IBD was defined according to the treating physician using accepted diagnostic criteria.

Indications for ERCP were in line with the EASL guidelines: presence of high grade or rapidly progressive strictures on imaging regardless of symptoms, or symptoms or biochemical signs of obstructive cholestasis and/or bacterial cholangitis. ERCP with brush cytology was performed by a predefined study management algorithm. Brush cytology and fluorescent in-situ hybridization (FISH) were performed during ERCP.

At yearly follow-ups, clinical data including symptoms, medication, endoscopic interventions, and the results of MRI/MRCPs were registered. Annual blood samples included liver function tests and CA19–9. The patient’s follow-up lasted 5 years or at the latest until 1 November 2023. The mean number of ERCPs was 2.6 (range 1–8).

On-demand ERCP

Patients under routine surveillance, which comprises of annual laboratory tests including CA19–9, Fibroscan, and ultrasound, at the Amsterdam University Medical Center were accrued from the EpiPSC2 registry [16]. The Amsterdam University Medical Center is a national expert center for PSC, although not a transplant center. Patients are frequently referred at an early stage of their disease for an expert opinion. An expected transplant-free survival of less than 1 year, and previous liver transplantation or hepatobiliary malignancy were not considered exclusion criteria.

Data were collected from the date of PSC diagnosis up to December 2023, liver transplantation, or death. Data were collected retrospectively from the date of diagnosis until 2008, and prospectively from that date onwards. The last day of follow-up was 31 December 2023. Routine annual MRI/MRCP was introduced in Amsterdam in 2023, so most of the patients had MRCI/MRCP imaging based on either their symptoms, an increase in cholestasis, their annual ultrasound findings, or a combination thereof. ERCP was performed according to the European Society of Gastrointestinal Endoscopy (ESGE)/EASL 2017 guidelines [9]. The mean number of ERCPs was four (range 0–45).

Overall cohort

In the final analysis, we included only cases where the PSC was diagnosed in 1990 or later, as MRCP came into use in the early 1990s. It became more widely adopted in clinical practice throughout the 1990s and, by the late 1990s and early 2000s, MRCP became a standard tool in many hospitals for diagnosing and surveillance of PSC. For the diagnostic strategies of CCA see Appendix 1s.

The final study cohort consisted of 1017 patients in the scheduled ERCP cohort, 412 patients in the annual MRI/MRCP cohort, and 200 patients in the on-demand ERCP cohort ([Fig. 1]). The composite end point consisted of hepatobiliary malignancy (CCA, HCC, or gallbladder cancer), transplantation, or liver-related death. The annual cumulative incidence of patients included in the different cohorts is shown in Fig. 3s.

Fig. 1 Flow chart of the study population. ERCP, endoscopic retrograde cholangiopancreatography; MRCP, magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging; PSC, primary sclerosing cholangitis.

Statistics

Summary statistics were described using mean and SD, median and interquartile range (IQR), or numbers and percentages. Statistical comparisons between groups were done using analysis of variance (ANOVA), the Kruskal–Wallis test, and chi-squared test.

Data from the three groups (scheduled ERCP, MRI/MRCP surveillance, and on-demand ERCP), as presented in [Table 1] and [Table 2], were analyzed. If a global significant P value was identified, post-hoc analysis with Sidak correction (significance level 0.05) was conducted. Median regression analysis (least-absolute-value models) was used to test for the change in median Fibrosis-4 screening score (FIB-4) between the groups. The incidence rate and incidence rate ratio (IRR) were calculated using Poisson regression models.

Table 1 Comparison of clinical characteristics and the first laboratory values available at the time of diagnosis in the different surveillance cohorts.

Variables	Scheduled ERCP (n = 1017)	MRI/MRCP surveillance (n = 412)	On-demand ERCP (n = 200)	P value [Multiple post-hoc comparisons]1
ALP, alkaline phosphatase; ALT, alanine transaminase; AOM, Amsterdam–Oxford PSC model; AST, aspartate transaminase; CA19–9, carbohydrate antigen 19–9; GGT, gamma-glutamyl transferase; Hb, hemoglobin; IBD, inflammatory bowel disease; INR, international normalized ratio. xULN, ratio to upper limit of normal. 1 Sidak’s multiple comparison procedure was used to correct significance levels for post-hoc testing (P < 0.05). 1/2, scheduled ERCP vs. MRI/MRCP surveillance; 1/3, scheduled ERCP vs. on-demand ERCP; 2/3, MRI/MRCP surveillance vs. on-demand ERCP.
Total person-years	11 621	7267	2482
Follow-up time, mean (SD), years	11 (7)	18 (7)	12 (7)
Sex, male, n (%)	591 (58)	281 (68)	126 (63)	0.002 [1/2]
Age at diagnosis, mean (SD), years	36 (14)	31 (14)	38 (15)	<0.001 [1/2, 1/3, 2/3]
IBD, n (%)	777 (76)	322 (78)	142 (71)	0.14
IBD diagnosed before PSC, n (%)	564 (73)	177 (55)	83 (58)	<0.001 [1/2, 1/3]
Hb, median (IQR), g/L	138 (127 to 148)	140 (131 to 149)	162 (145 to 175)	<0.001 [1/2, 1/3, 2/3]
Platelets, median (IQR), 109/L	264 (216 to 320)	245 (197 to 304)	270 (219 to 356)	<0.001 [1/2, 2/3]
AST, median (IQR), U/L	36 (26 to 57)	40 (27 to 81)	48 (31 to 95)	<0.001 [1/2, 1/3, 2/3]
xULN female	0.91 (0.66 to 1.37)	0.96 (0.67 to 1.59)	1.29 (0.84 to 2.21)	0.10
xULN male	0.87 (0.60 to 1.40)	0.96 (0.64 to 2.03)	1.23 (0.81 to 2.23)	<0.001 [1/2, 1/3]
ALT, median (IQR), U/L	44 (25 to 87)	50 (28 to 120)	73 (36 to 140)	<0.001 [1/2, 1/3, 2/3]
xULN female	0.97 (0.60 to 1.89)	0.90 (0.43 to 1.74)	2.21 (0.88 to 2.88)	0.002 [1/2, 1/3, 2/3]
xULN male	1.06 (0.58 to 2.04)	0.86 (0.46 to 1.98)	1.66 (0.91 to 3.41)	<0.001 [1/3, 2/3]
ALP, median (IQR), U/L	134 (96 to 220)	133 (78 to 289)	221 (128 to 323)	<0.001 [1/3,2/3]
xULN female	1.19 (0.82 to 1.81)	1.00 (0.58 to 2.17)	2.30 (1.16 to 3.01)	<0.001 [1/3, 2/3]
xULN male	1.34 (0.97 to 2.37)	1.22 (0.74 to 2.64)	1.54 (1.09 to 3.17)	0.007 [1/2, 2/3]
GGT, median (IQR), U/L	103 (38 to 256)	127 (33 to 380)	190 (81 to 347)	<0.001 [1/2, 1/3, 2/3]
xULN female	1.73 (0.73 to 4.25)	1.26 (0.30 to 3.77)	2.51 (1.34 to 4.46)	<0.001 [1/2, 2/3]
xULN male	2.27 (0.73 to 5.40)	1.37 (0.38 to 3.78)	1.98 (1.08 to 3.39)	<0.001 [1/2]
Total bilirubin, median (IQR), µmol/L	11.0 (8.0 to 17.0)	11.0 (8.0 to 17.0)	13.0 (8.0 to 28.0)	0.02 [1/3, 2/3]
xULN, female	0.45 (0.35 to 0.70)	0.40 (0.24 to 0.52)	0.63 (0.32 to 0.95)	<0.001 [1/3, 2/3]
xULN male	0.60 (0.40 to 0.95)	0.65 (0.43 to 0.95)	0.54 (0.46 to 0.75)	0.55
Albumin, median (IQR), g/L	38.0 (35.4 to 40.9)	39.0 (37.0 to 42.0)	42.0 (39.0 to 45.0)	<0.001 [1/2, 1/3, 2/3]
INR, median (IQR)	1.0 (0.9 to 1.1)	1.0 (0.9 to 1.1)	1.0 (1.0 to 1.1)	0.14
CA19–9, median (IQR), U/L	5.0 (3.0 to 10.0)	10.0 (4.5 to 19.0)	12.0 (9.0 to 39.0)	<0.001 [1/2, 1/3, 2/3]
Mayo Score, median (IQR)	−0.42 (−0.90 to 0.09)	−0.56 (−1.09 to 0.08)	−0.41 (−1.13 to 0.46)	0.02 [1/2]
FIB-4, median (IQR)	0.74 (0.48 to 1.17)	0.73 (0.46 to 1.16)	0.76 (0.53 to 1.20)	0.59
AOM, median (IQR)	1.49 (1.17 to 1.86)	1.51 (1.11 to 1.89)	1.60 (1.20 to 1.89)	0.36

Table 2 Comparison of the incidences – cumulative and per 1000 patient-years (95%CI) – for the composite end point (liver and biliary malignancy, liver transplantation, and death) and the individual components during 30 years of follow-up.

	Scheduled ERCP (n = 1017)	MRI/MRCP surveillance (n = 412)	On-demand ERCP (n = 200)	P value [Post-hoc]1
1 Adjusted P value (for sex and age at diagnosis) between the groups with Sidak’s multiple comparison procedure was used to correct significance levels for post-hoc testing (P < 0.05). 1/2, scheduled ERCP vs. MRI/MRCP surveillance; 1/3, scheduled ERCP vs. on-demand ERCP; 2/3, MRI/MRCP surveillance vs. on-demand ERCP. 2 Competing cause of nonliver-related deaths. 3 Competing cause of other outcomes.
Composite end point
Number of events	143	94	70
Cumulative2, %	14.1 (12.0–16.4)	22.8 (18.8–27.2)	35.0 (28.4–42.0)	<0.001 [1/2, 1/3, 2/3]
Per 1000 patient-years	12.3 (10.4–14.5)	13.0 (10.5–15.9)	28.2 (22.0–35.6)	<0.001 [1/3, 2/3]
Cholangiocarcinoma
Number of events	31	16	19
Cumulative3, %	6.7 (3.8–10.6)	8.8 (4.5–14.8)	26.9 (13.2–42.7)	0.003 [1/3, 2/3]
Per 1000 patient-years	2.7 (1.8–3.8	2.2 (1.3- 3.6)	7.7 (4.6–12.0)	<0.001 [1/3, 2/3]
Hepatocellular carcinoma
Number of events	8	2	4
Cumulative3, %	1.8 (0.8–2.8)	0.8 (0.1–2.9)	4.8 (1.7–11.2)	0.37
Per 1000 patient-years	0.7 (0.3–1.4)	0.3 (0.0–1.0)	1.6 (0.4 -4.1)	0.29
Gallbladder cancer
Number of events	7	3	3
Cumulative3, %	2.7 (1.0–6.0)	1.5 (0.3–0.5)	2.2 (0.6–5.9)	0.58
Per 1000 patient years	0.6 (0.2–1.2)	0.4 (0.1–1.2)	1.2 (0.2- 3.5)	0.54
Liver transplantation
Number of events	96	70	40
Cumulative3, %	29.5 (22.3–36.9)	31.2 (23.2–39.5)	39.7 (27.2–52.0)	0.005 [1/3, 2/3]
Per 1000 patient-years	8.3 (6.7–10.1)	9.7 (7.5–12.2)	16.1 (11.5–21.9)	0.001 [1/3, 2/3]
Liver-related death
Number of events	6	7	5
Cumulative3, %	2.5 (0.7–6.5)	3.3 (1.1–7.5)	5.0 (1.8–7.5)	0.20
Per 1000 patient-years	0.5 (0.2–1.1)	1.0 (0.4–2.0)	2.0 (0.7–4.7)	0.12
Liver transplantation for suspected malignancy
Number of events	29	11	NA
Cumulative3, %	2.9 (1.9–4.0)	2.7 (1.3–4.7)	NA	0.84
Per 1000 patient-years	2.5 (1.7–3.6)	1.5 (0.8–2.7)	NA	0.27
Nonliver-related death
Number of events	19	4	7
Cumulative, %	1.9 (1.1)	1.0 (0.3–2.5)	3.5 (1.4–7.0)	0.09
Per 1000 patent-years	1.6 (1.0–2.6)	0.6 (0.2–1.4)	2.8 (1.1–5.8)	0.24

For cumulative incidence functions, we used Fine and Gray competing-risks regression models (robust estimate of variance) and estimated adjusted subhazard ratios (sHRs). The major difference between the Fine and Gray model and classic time-to-event models is that the risk set for the end point of interest includes not only those who have not yet failed because of the primary end point but also those who have failed because of a competing cause (causes of nonliver-related deaths). This model adjusts for age and sex. Poisson models goodness-of-fit were assessed using deviance and Pearson chi-squared tests, which showed a good fit in all models. Age and sex were used as covariates in these models.

A possible nonlinear relationship between the composite end point and calendar years was assessed using 4-knot restricted cubic spline Poisson regression models, with the knots located at the 5th, 35th, 65th, and 95th percentiles. For restricted cubic splines, also known as natural splines, knot locations were based on Harrell’s recommended percentiles. The normality of variables was evaluated graphically and using the Shapiro–Wilk W test.

Stata 17.0 (StataCorp LP, College Station, Texas, USA) was used for the statistical analyses.

Ethics

Scheduled ERCP

All the patients included in the PSC registry provided written informed consent. The study was performed following the principles of the good clinical practice and in accordance with the ethical guidelines of the Declaration of Helsinki (6th revision, 2008). The study protocol was approved by the Helsinki University Hospital Ethical Committee IV (HUS/1566/2020).

Annual MRI/MRCP surveillance

The regional Ethics Committee in Stockholm approved the study (Dnr 2011/824–31/2). All patients included provided written informed consent.

On-demand ERCP

The study was approved by the institutional review board of UMC Utrecht (reference: NL14614.041.06/METC 06–267/E).

Results

Characteristics of the study cohorts

The final study cohort consisted of 1629 patients. The clinical characteristics of the patients and laboratory findings at the time the first laboratory values were available are shown in [Table 1]. The total number of person-years was 11621, 7267, and 2482 for scheduled ERCP, MRI/MRCP surveillance, and on-demand ERCP, respectively. The mean follow-up time varied from 11 to 18 years. Age at diagnosis and levels of liver enzymes were higher in the on-demand ERCP group compared with the other cohorts; however, the FIB-4 and the Amsterdam–Oxford PSC Model (AOM) scores did not differ between the cohorts. Alkaline phosphatase (ALP) levels were elevated >1.5× the upper limit of normal in 40.8%, 40.3%, and 57.4% of patients in the scheduled ERCP, MRI/MRCP surveillance, and on demand ERCP cohorts, respectively. The progression of fibrosis, as assessed by the change in FIB-4 from baseline to the end of follow-up (ΔFIB-4), was lowest in the scheduled ERCP cohort and most pronounced in the MRI/MRCP group (Fig. 4s).

Outcomes of the surveillance

We analyzed the cumulative incidence of CCA, HCC, gallbladder cancer, liver transplantation, liver-related death, and incidence per 1000 person-years in the three different cohorts ([Table 2]). The cumulative incidence of the composite end point was lowest in the of scheduled ERCP group (14.1%, 95%CI 12.0%–16.4%) and highest in the on-demand ERCP cohort (35.0%, 95%CI 28.4%–42.0%). The cumulative incidence of patients reaching the composite end point for each of the cohorts is shown in [Fig. 2], showing that the overall prognosis was best in the scheduled ERCP cohort.

Fig. 2 Cumulative incidence of patients reaching the composite end point in the three cohorts (competing causes of nonliver-related death). ERCP, endoscopic retrograde cholangiopancreatography; MRCP, magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging.

The analysis of the cumulative incidence of CCA in the three cohorts shows that the incidence was highest in the on-demand ERCP cohort ([Fig. 3]). The incidence was initially higher in the scheduled ERCP cohort with systematic brush cytology than in the MRI/MRCP surveillance group; however, during the follow-up period, the incidence increased in the MRI/MRCP surveillance group, whereas in the scheduled ERCP cohort, the increase was less pronounced.

Fig. 3 Cumulative incidence of: a patients diagnosed with cholangiocarcinoma (CCA); b CCA and liver transplantation (LTx) done for suspicion of biliary malignancy. ERCP, endoscopic retrograde cholangiopancreatography; MRCP, magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging.

We also analyzed the cumulative incidence of liver transplantation performed for suspicion of CCA ([Fig. 3] b). Data on the indications for liver transplantation were available only for the scheduled ERCP and MRI/MRCP surveillance cohorts. The cumulative incidence was higher in the scheduled ERCP cohort, as having repeatedly documented suspicion of malignancy on brush cytology was an indication for evaluation for liver transplantation (Fig. 2s).

The cumulative incidence of CCA was also lowest in the scheduled ERCP group compared with the other groups, but it did not differ statistically significantly from the MRI/MRCP surveillance group. The incidence of HCC/1000 patient-years was markedly lower in the MRI/MRCP surveillance group than in the other cohorts. The incidence of liver transplantation did not differ between the scheduled ERCP and MRI/MRCP surveillance groups, but was significantly higher in the on-demand ERCP group. The incidence of death, both liver related and nonliver related, was similar across all cohorts.

Finally, we evaluated the time trends in reaching the composite end point in the different surveillance strategy cohorts from 1990 to 2023 ([Fig. 3]). There was a clear reduction in the incidence of the composite end point, especially in the scheduled ERCP cohort but also in the on-demand ERCP cohort. In contrast, no reduction was seen in the annual MRI/MRCP surveillance cohort.

Discussion

In this study, we compared three PSC patient cohorts with different diagnosis and surveillance strategies – scheduled ERCP, annual MRI/MRCP surveillance, and on-demand ERCP – and the impact of these on their outcomes. Scheduled ERCP surveillance was associated with the lowest fibrosis progression based on ΔFIB-4 and an improved overall prognosis, as assessed by a composite end point of hepatobiliary malignancy, liver transplantation, and liver-related death. Moreover, there was a significant reduction over time in the incidence of the composite end point in the scheduled ERCP cohort compared with the MRI/MRCP surveillance group. The cohort following an on-demand ERCP follow-up strategy, as per the current EASL recommendations, exhibited the worst prognosis and, although there was an improving trend toward reaching the composite end point, the incidence per 1000 person-years remained above the mean for the entire population ([Fig. 4]).

Fig. 4 Time trends for reaching the composite end point with the different surveillance strategies. Relationships between the composite end point and calendar years were assessed using 4-knot restricted cubic spline Poisson regression models. Shaded areas indicate the 95%CIs; dotted line shows the mean incidence of the composite end point/1000 patient years in the whole study population. ERCP, endoscopic retrograde cholangiopancreatography; MRCP, magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging.

Regular imaging surveillance of PSC patients for hepatobiliary cancers has previously been shown to be associated with better survival [17] and lower risk of hepatobiliary cancer-related death, but the study could not show a difference in post-CCA survivorship, specifically, between the surveillance and nonsurveillance groups. A British study evaluating the effect of annual imaging on the risk of hepatopancreaticobiliary (HPB) cancer-related death demonstrated that imaging surveillance was associated with >2-fold risk reduction in HPB cancer-related death (HR 0.43, 95%CI 0.23–0.80; P = 0.04); however, after the exclusion of all CCA cases from the first year, there was no difference in post-CCA survivorship [18].

Improved surveillance strategies are urgently needed [11] [19]; however, surveillance of PSC, including the progression of bile duct changes, the risk of hepatobiliary malignancy, and the development of cirrhosis, is challenging. At present, there are very scarce data on markers for predicting disease activity and the development of new strictures. Most surveillance recommendations are based on expert opinions and have only weak study evidence to support them. Surveillance based on scheduled ERCP enables diagnosis of CCA, and early detection and treatment of relevant strictures to salvage liver function. Clinical practice guidelines vary with regard to surveillance of PSC-associated hepatobiliary malignancy, partly owing to the paucity of data regarding the impact of surveillance on clinical outcomes [20]. Recently, three clinical guidelines have been published [9] [20] [21] (Table 1s). For hepatobiliary malignancy, both the EASL and American Association for the Study of Liver Diseases (AASLD) recommend annual surveillance with ultrasound and/or MRI/MRCP in patients with large duct disease.

An Australian multicenter study including 298 PSC patients, of whom 74% participated in regular MRCP surveillance, showed a 71% reduction of the risk of death and an increased likelihood of having an earlier ERCP after PSC diagnosis in patients with a dominant stricture; however, survival after post-hepatobiliary cancer diagnosis was not significantly different between the surveillance and nonsurveillance groups [22]. This is in line with the study of Trivedi et al. [18] and underlines the importance of even earlier detection of CCA, at the stage of biliary dysplasia. ERCP with biliary biopsy and brushings is recommended by the AASLD in PSC patients with a relevant stricture without a mass on MRI/MRCP [20]. In the British guidelines, MRCP or contrast-enhanced computed tomography should be performed based on symptom monitoring: new or changing symptoms, or evolving abnormalities on laboratory investigations. However, it has previously been shown that, even in mostly asymptomatic PSC patients, 43% already had advanced bile duct disease, and 7% presented with suspicious or malignant brush cytology at their first ERCP [12]. Up to 25% of CCA patients are asymptomatic at the time of diagnosis [11]. A recent study suggests only moderate agreement between ERCP and MRI/MRCP in assessing the presence and severity of biliary strictures [23].

In a large retrospective study, including 2975 PSC patients, that evaluated different follow-up strategies, the HR for death, compared with no surveillance, was 0.64 (95%CI 0.48–0.86) for ultrasound/MRI surveillance and 0.53 (95%CI 0.37–0.75) including scheduled ERCP [11]. Hepatobiliary malignancy was diagnosed in 5.9% of the patients at 7.9 years of follow-up, which is in line with our scheduled ERCP and annual MRI/MRCP groups, but significantly lower than in the on-demand ERCP cohort. In a retrospective study (n = 286), scheduled ERCP showed a better transplantation-free survival rate than on-demand ERCP for PSC patients with a dominant stricture (51% vs. 29%). This was regardless of symptoms or whether the dominant stricture was diagnosed during surveillance or at the time of PSC diagnosis [24]. There is increasing evidence for the more liberal use of ERCP, including from the data of Bergquist et al. [11].

In the present study, surveillance with scheduled ERCP with brush cytology was associated with an improved overall survival, even improving over time, but interestingly the cumulative incidence of CCA and incidence per 1000 patient-years did not differ statistically significantly between the scheduled ERCP and annual MRI/MRCP surveillance groups, being markedly lower than in the on-demand ERCP group (P < 0.001). The incidence of liver transplantation owing to suspicion of malignancy also did not differ between the scheduled ERCP and annual MRI/MRCP surveillance groups. In the scheduled ERCP cohort, all of the patients were followed from diagnosis, whereas in the MRI/MRCP surveillance cohort, patients with hepatobiliary malignancies were initially excluded, before their inclusion into the surveillance program from 1 November 2011, and the cohort is not included in the present study. Endoscopic surveillance with brush cytology aims to detect biliary dysplasia before the development of CCA, whereas MRI/MRCP surveillance is based on the detection of a suspicious CCA mass early enough for further interventions. Early detection of biliary neoplasia, before the development of CCA, is of great importance to allow for curative treatment, such as prophylactic liver transplantation or surgical resection. In most European centers, documented CCA is a contraindication for liver transplantation.

The improved overall survival of patients undergoing scheduled ERCP may also be associated with the dilation of strictures, which has been shown to be associated with improved prognosis [23]. The study of Bergquist et al. demonstrated nearly a two-fold reduction in mortality among patients undergoing annual surveillance including ERCP [11] but, as in the present study, the risk of death from CCA was not decreased. The incidence of HCC was lower in the annual MRI/MRCP cohort than in the other cohorts, with an incidence of 0.3%, but still in line with previous studies, where the lifetime incidence varied between 0.3% and 2.8% [7]. In PSC patients, HCC was present only in patients with cirrhosis [17]. Most HCC cases are incidentally detected at the time of liver transplantation [7].

The strength of this study is its prospective collection of data and the large cohort sizes with long-term follow-up. To our knowledge, this is the first study comparing systematic ERCP-based surveillance with the current clinical guidelines, using detailed outcome data of hepatobiliary malignancies, transplantation, indications for transplantation, and liver-related and nonliver-related deaths.

As in all registry studies, there is always a risk of not including all PSC patients. Selection bias with different cohorts from the different centers may have influenced the results and real-life adherence to annual surveillance is not 100% [10]. In addition, the exclusion of some patients with transplantation and hepatobiliary malignancy before inclusion has an impact on the comparison of the MRI/MRCP cohort and the scheduled ERCP cohort. The data should therefore be interpreted with caution. In addition, MRIs in both the annual MRI/MRCP and scheduled ERCP cohorts were done in the patient’s local university hospitals. It should be noted that ERCP is an invasive method and carries a risk of complications, most commonly pancreatitis. In this study, we did not have data on the complications associated with ERCP but, in our recent series including mild cases too, the overall risk for pancreatitis was 5.7% and for post-ERCP cholangitis was 1.5%–2.8% [25].

In conclusion, this large multicenter study demonstrates that a surveillance strategy based on scheduled ERCP and individual risk stratification is associated with an improved overall prognosis and a significantly decreased cumulative incidence of CCA compared with on-demand ERCP (as recommended in current clinical guidelines). This reduction is attributed to systematic screening for biliary neoplasia using brush cytology and timely liver transplantation before the development of CCA. Additionally, we have shown that this strategy is associated with the lowest rate of fibrosis progression and improved outcomes over time.

Conflict of Interest

The authors declare that they have no conflict of interest.

Acknowledgement

For patient data collection, the Swedish Hepatology Research Network (SWEHEP) is acknowledged, including Nilsson Emma, Mårten Werner, Stergios Kechagias, Nils Nyhlin, Annika Bergquist, Antonio Molinaro, Johan Vessby, Christina Villard.

• References

• 1 Ponsioen CY, Assis DN, Boberg KM. et al. Defining primary sclerosing cholangitis: results from an International Primary Sclerosing Cholangitis Study group consensus process. Gastroenterology 2021; 161: 1764-1775
• 2 Bergquist A, Ekbom A, Olsson R. et al. Hepatic and extrahepatic malignancies in primary sclerosing cholangitis. J Hepatol 2002; 36: 321-327
• 3 Manninen P, Karvonen AL, Laukkarinen J. et al. Colorectal cancer and cholangiocarcinoma in patients with primary sclerosing cholangitis and inflammatory bowel disease. Scand J Gastroenterol 2015; 50: 423-428
• 4 Boonstra K, Weersma RK, van Erpecum KJ. et al. Population-based epidemiology, malignancy risk, and outcome of primary sclerosing cholangitis. Hepatology 2013; 58: 2045-2055
• 5 Barner-Rasmussen N, Pukkala E, Jussila A. et al. Epidemiology, risk of malignancy and patient survival in primary sclerosing cholangitis: a population-based study in Finland. Scand J Gastroenterol 2020; 55: 74-81
• 6 Fung BM, Lindor KD, Tabibian JH. Cancer risk in primary sclerosing cholangitis: Epidemiology, prevention, and surveillance strategies. World J Gastroenterol 2019; 25: 659-671
• 7 Zenouzi R, Weismüller TJ, Hübener P. et al. Low risk of hepatocellular carcinoma in patients with primary sclerosing cholangitis with cirrhosis. Clin Gastroenterol Hepatol 2014; 12: 1733-1738
• 8 Karlsen TH, Folseraas T, Thorburn D. et al. Primary sclerosing cholangitis - a comprehensive review. J Hepatol 2017; 67: 1298-1323
• 9 EASL. Clinical Practice Guidelines on sclerosing cholangitis. J Hepatol 2022; 77: 761-806
• 10 Villard C, Friis-Liby I, Rorsman F. et al. Prospective surveillance for cholangiocarcinoma in unselected individuals with primary sclerosing cholangitis. J Hepatol 2023; 78: 604-613
• 11 Bergquist A, Weismüller TJ, Levy C. et al. Impact on follow-up strategies in patients with primary sclerosing cholangitis. Liver Int 2023; 43: 127-138
• 12 Boyd S, Tenca A, Jokelainen K. et al. Screening primary sclerosing cholangitis and biliary dysplasia with endoscopic retrograde cholangiography and brush cytology: risk factors for biliary neoplasia. Endoscopy 2016; 48: 432-439
• 13 Stiehl A, Rudolph G, Klöters-Plachky PC. et al. Development of dominant bile duct stenoses in patients with primary sclerosing cholangitis treated with ursodeoxycholic acid: outcome after endoscopic treatment. J Hepatol 2002; 36: 151-156
• 14 Boyd S, Mustamäki T, Sjöblom N. et al. NGS of brush cytology samples improves the detection of high-grade dysplasia and cholangiocarcinoma in patients with primary sclerosing cholangitis: A retrospective and prospective study. Hepatol Commun 2024; 8: e0415
• 15 Barner-Rasmussen N, Sjöblom N, Arola J. et al. The role of serology, liver function tests and imaging in screening of primary sclerosing cholangitis: the HelPSCreen score. Scand J Gastroenterol 2023; 58: 1491-1498
• 16 van Munster KN, Mol B, Goet JC. et al. Disease burden in primary sclerosing cholangitis in the Netherlands: A long-term follow-up study. Liver Int 2023; 43: 639-648
• 17 Ali AH, Tabibian JH, Nasser-Ghodsi N. et al. Surveillance for hepatobiliary cancers in patients with primary sclerosing cholangitis. Hepatology 2018; 67: 2338-2351
• 18 Trivedi PJ, Crothers H, Mytton J. et al. Effects of primary sclerosing cholangitis on risks of cancer and death in people with inflammatory bowel disease, based on sex, race, and age. Gastroenterology 2020; 159: 915-928
• 19 Weismüller T, Trivedi PJ, Bergquist A. et al. Patient age, sex, and inflammatory bowel disease phenotype associate with course of primary sclerosing cholangitis. Gastroenterology 2017; 152: 1975-1984
• 20 Bowlus CL, Arrivé L, Bergquist A. et al. AASLD practice guidance on primary sclerosing cholangitis and cholangiocarcinoma. Hepatology 2023; 77: 659-702
• 21 Chapman MH, Thorburn D, Hirschfield GM. et al. British Society of Gastroenterology and UK-PSC guidelines for the diagnosis and management of primary sclerosing cholangitis. Gut 2019; 68: 1356-1378
• 22 Tan N, Ngu N, Worland T. et al. Surveillance MRI is associated with improved survival in patients with primary sclerosing cholangitis. Hepatol Commun 2024; 8: e0442
• 23 Tenca A, Mustonen H, Lind K. et al. The role of magnetic resonance imaging and endoscopic retrograde cholangiography in the evaluation of disease activity and severity in primary sclerosing cholangitis. Liver Int 2018; 38: 2329-2339
• 24 Rupp C, Hippchen T, Bruckner T. et al. Effect of scheduled endoscopic dilatation of dominant strictures on outcome in patients with primary sclerosing cholangitis. Gut 2019; 68: 2170-2178
• 25 Koskensalo V, Aronen P, Färkkilä M. et al. Use of thiopurines is not a risk factor for post- ERCP pancreatitis in patients with primary sclerosing cholangitis. Dig Liver Dis 2021; 53: 1020-1027

Publication History

Received: 30 August 2024

Accepted after revision: 13 December 2024

Article published online:
28 January 2025