Epidemiology and outcomes of primary sclerosing cholangitis: an Australian multicentre retrospective cohort study

We performed a retrospective cohort study of patients diagnosed with PSC from 1st January 2000 to 30th March 2021 at 9 tertiary academic liver centers across Australia in Victoria, New South Wales, and South Australia, including two transplant centers and seven non-transplant tertiary referral centers.

We included all adult patients with a confirmed diagnosis of PSC based on the presence of established cholangiographic or histological features of large or small duct PSC coupled with a cholestatic biochemical profile. Patients with secondary causes of sclerosing cholangitis as per recognized international guidelines were excluded [8‐10]. Patients with concomitant liver diseases were not excluded for the purpose of this study.

Case ascertainment was achieved at a local level with a combination of pre-existing or prospectively collected databases in addition to computerized search of patient diagnoses through electronic medical records, radiology, and histology records using the terms—“primary sclerosing cholangitis” and “sclerosing cholangitis”. Pre- and post-liver transplant patient registries were also reviewed at the two transplant centers. All individual medical records were reviewed at a local level to confirm a diagnosis of PSC and accuracy of data collection, including removal of duplicate patient entries when referred for LT.

Clinical demographics and characteristics collected included patient sex, date of birth, date of PSC diagnosis, ethnicity, phenotype, ursodeoxycholic acid (UDCA) use, IBD type, presence of cirrhosis with or without portal hypertension, decompensation events (development of ascites, hepatic encephalopathy, or variceal bleeding), LT indication and date, malignancy type and date of diagnosis.

Evidence for overlap syndrome with autoimmune hepatitis was based on biochemical or serological features with or without histological confirmation. IBD phenotypes of Crohn’s disease, ulcerative colitis or IBD-unclassified (IBDU) were determined via clinical documentation and biochemical, imaging, endoscopic and histological confirmation if available based on consensus guidelines [11]. The presence of cirrhosis and portal hypertension was based on radiological or histological findings if available coupled with supportive clinical, non-invasive testing with transient elastography and laboratory data [12].

Standardized incidence (SIR) and mortality ratios (SMR) were calculated with the indirect method using population data from the Australian Institute of Health and Welfare as the reference population. Relative survival is defined as the ratio of the observed all-cause survival among patients and the expected survival in a comparable group in the general population. Life tables for the general Australian population stratified by calendar year, age and sex were used to estimate the relative survival. Observed survival in individuals with PSC was calculated on 2-yearly intervals from the date of PSC diagnosis and compared to age and sex matched data from population life tables as described previously [13].

A total of 413 PSC patients were included in the study with a median follow-up time of 7.1 (IQR 3.6–11.7) years, contributing to 3,285 person-years of follow-up time from diagnosis. The median age at PSC diagnosis was 37 (IQR 23–53) years, and 266 (64%) patients were male with the majority (88%) of patients being of European ancestry. Two hundred and nineteen (53%) patients were initially managed at a liver transplant tertiary center. Three hundred and seventy-one (90%) patients had a large duct phenotype at diagnosis, and 294 (71%) had associated IBD. Three (0.7%) patients had concomitant chronic hepatitis C. Two hundred and sixty-four (64%) patients were taking ursodeoxycholic acid (UDCA) at any stage during their follow-up. Further descriptions of the demographics and clinical characteristics of the overall cohort are presented in Table 1.

The annual number of new PSC cases at the contributing centers in greater Melbourne increased from 7 in 2000 to a peak of 25 cases in 2017 but declined to 5 toward the end of the study period in 2020/2021. This corresponded to an average annual incidence of 0.27 cases per 100,000 person-year in greater Melbourne (Fig. 1). Towards the end of the study in 2019/2020, there was a total of 266 cases (165 males, 101 females) of PSC alive and managed in the contributing hospitals in greater Melbourne corresponding to an estimated prevalence of 5.9 cases per 100,000 inhabitants (males; 3.7:100,000, females; 2.2:100,000).

This is the largest published study of a Southern Hemisphere cohort of patients with PSC and demonstrates that the natural history of PSC patients in Australia is similar to that of their Northern Hemisphere counterparts. PSC remains an uncommon disease in Australia (266 patients in 2019/2020 managed in 7 Victorian hospitals with a catchment population of ~ 4.5 million). Thirty percent of the cohort required liver transplant or died at a median time from diagnosis of 13 years, with development of cirrhosis, CCA and increased age at PSC diagnosis being significantly associated with reduced transplant-free survival. Despite the increased risk of developing hepatobiliary and colorectal malignancy, only PSC-related CCA was associated with an over 240-fold increase in mortality compared to the general population with a median survival of only 7 months. The presence of a dominant stricture, history of colectomy and increased age at PSC diagnosis was significantly associated with increased risk of CCA development. Furthermore, we demonstrate a reduction in relative survival with increased age and time from diagnosis as compared to the Australian general population.

Our cohort draws similarities to other European populations with the majority being of male gender with concomitant IBD and a median age of diagnosis at 37 years [3, 14‐16]. Our estimated prevalence rate of 5.9 cases per 100,000 inhabitants is similar to the highest prevalence reported in a United Kingdom study [15]; however, this was significantly lower than the point prevalence of 11 cases per 100,000 inhabitants in Canterbury, New Zealand [7]. This difference in prevalence despite geographical proximity may be due to complete identification of all PSC patients under the care of a small group of gastroenterologists in Canterbury, reflecting a true population prevalence rate [7]. An alternative explanation would be an increased burden of PSC in association with predisposition towards autoimmune disease at higher latitudes [17].

Forty-one percent of our study cohort were cirrhotic, with 8% diagnosed with cirrhosis at time of PSC diagnosis. Patients who were cirrhotic at diagnosis were less likely to have concomitant IBD, which may reflect regular screening with liver function tests prompting earlier diagnosis in this cohort. Majority of cirrhotic patients had concomitant portal hypertension and almost 50% experienced at least one episode of hepatic decompensation. This seems to be higher compared to other cholestatic liver conditions such as primary biliary cholangitis, where less than 30% of patients develop liver failure 10 years after diagnosis [18]. Perhaps related to the lack of an effective medical therapy, PSC patients seem to develop cirrhosis and liver decompensation at higher rates compared to other liver conditions. The finding of increased proportion of cirrhosis in patients on UDCA therapy likely reflects a degree of selection bias rather than true influence of UDCA on outcomes, with patients of a more severe phenotype and/or higher ALP levels being more likely to be commenced on therapy. This may also explain the increased proportion of deaths in patients who took UDCA compared to those who did not.

The proportion of PSC patients that develop cirrhosis has been reported to develop at an incidence rate of 18.6 per 1000 person-years in a American population-based study [15], which is significantly lower than our incidence rate of 4 per 100 person-years. This may reflect an element of referral bias, as half of our patients were managed primarily at transplant centers. Also, the increased risk of developing cirrhosis early after diagnosis (Supp Fig. 2: Risk of developing cirrhosis over time) may reflect latent disease that had been undiagnosed for years in asymptomatic patients or be representative of patients with an aggressive disease phenotype.

CCA remains a significant cause of morbidity and mortality in PSC patients [2, 19, 20]. The proportion of patients who developed CCA (5.8%) is comparable to other studies with prevalence ranging from 2.5 to 12% [2, 14, 21, 22]. The median time from diagnosis to development of CCA in our cohort was 6.7 years, which mirrors the findings of a large population-based study in the Netherlands where median time to CCA was 6 years [20]. This is longer compared to previous studies where most cases of CCA were detected in the first year of PSC diagnosis [14, 20, 22]. The risk of CCA and CRC development compared to the general population are similar to other European cohorts at 240-fold and fivefold, respectively [14, 20]. Our findings of the presence of a dominant stricture and increased age of PSC diagnosis being risk factors for development of CCA have been previously reported in the literature [23, 24]. The increased association of CCA with colectomy has been demonstrated in other studies along with presence of IBD [25, 26], although the latter was not replicated in our findings. It is unclear the mechanisms that colectomy may confer increased risk of CCA, but this finding lends weight to the argument that the gut microbiome plays a pathogenetic role in disease development in PSC.

The reported transplant-free survival of 13 years in our Australian cohort is comparable to the largest international PSC cohort ever described—derived from 35 tertiary centers of over 7000 patients [19]. This also matched the estimated median transplant-free survival in transplant cohorts described by Boonstra et al. [20], of which the study also highlighted the greatly improved median transplant-free survival in the population-based cohort of 21.3 years in comparison to 13.2 years in the transplant center cohort. Increased age at PSC diagnosis, development of cirrhosis and CCA has also been found to be significant predictors of transplant-free survival in keeping with the current literature [21, 23, 27].

Another important observation from our study was that patients who are diagnosed at an older age do poorly compared to their same-aged counterparts in the general population. This effect is even seen in younger age groups, although it is more pronounced as patients are diagnosed at increasing age especially after 60 years. Overall, our findings do support the current literature that age of diagnosis has a significant impact on disease course [16, 19, 21], possibly related to an unique disease phenotype [23].

Our study highlights the difficulty and importance of diagnosing PSC early and suggests a role for screening higher risk individuals such as middle-aged males of European descent, particularly those with IBD. In addition, non-smokers may be at increased risk of developing PSC [28], including those with first-degree relatives with PSC [29]. Once the diagnosis of PSC is established, it is prudent to monitor for development of complications such as cirrhosis and CCA especially in older patients. Although there is lacking evidence on the benefit of surveillance of hepatobiliary malignancy in PSC, this is strongly recommended by international guidelines [9, 30] and screening (especially for CCA) may be warranted in all patients but perhaps more frequently in those with risk factors such as diagnosed at an older age, presence of a dominant stricture or history of colectomy.

Our study strengths include being the largest multicentre study of PSC in the Southern Hemisphere and long duration of patient follow-up. We also included comprehensive clinical endpoints such as the development of cirrhosis, malignancy, survival and need for LT. This was compared to the general population to derive the impact of PSC-related complications on overall morbidity and mortality. Identified limitations include our study being a retrospective analysis, as we acknowledge the real-world challenges of complete data collection and potential bias in case acquisition. Clinical delays with PSC diagnosis due to the typical asymptomatic presentation of PSC [1] may have affected results of the survival analysis. This is a limitation with most retrospective studies on PSC as we cannot accurately determine its subclinical course. However, our study reflects real-world data that correlates with other PSC cohorts as previously described. Although we did not exclude patients with other liver diseases, this contributed to less than 1% of the cohort and by including these patients allowed for a greater estimate of the epidemiology of PSC in Victoria.

We estimated the prevalence and incidence of a Victorian tertiary cohort of PSC patients in greater Melbourne; however, this does not reflect a true population-based result due to possible incomplete case acquisition and referrals from other states (Tasmania) to our Victorian liver transplant center contributing to a wider catchment population than expected. We also possibly missed out on a small subgroup of patients who may be managed privately or regionally that may have milder phenotypes of disease with improved outcomes. We acknowledge that some patients in our cohort were included in previous Australian studies [5, 6].

This study describes the demographics, clinical characteristics, and outcomes of the largest ever reported Australian cohort of PSC patients. We demonstrate a reduction in relative survival in patients diagnosed with older age with increased time from diagnosis, coupled with increased morbidity and mortality from development of CCA and cirrhosis. In addition, we found that presence of a dominant stricture and previous history of colectomy were risk factors for CCA.

Further prospective studies on PSC in Australia encompassing clinical data and serial prognostic markers are underway, which will allow us to further stratify high-risk patient groups and improve our understanding and management of PSC patients.