Primary sclerosing cholangitis (PSC) is a chronic inflammatory disease that causes fibrosis of the biliary tree. Life expectancy of patients is reduced by liver failure and a high incidence of malignancy. It is closely associated with inflammatory bowel disease, particularly ulcerative colitis, which coexists in approximately three-quarters of northern European patients. Cancers include cholangiocarcinoma, gallbladder cancer, hepatocellular carcinoma, pancreatic cancer and colorectal cancer. Ursodeoxycholic acid appears to reduce the incidence of colorectal neoplasia in patients with PSC, and there is some suggestion that it may also reduce the incidence of cholangiocarcinoma. A chemoprotective benefit of 5-aminosalicylates has not been confirmed in patients with PSC with associated inflammatory bowel disease. There is no accepted screening programme for cholangiocarcinoma, but methods for detecting early disease using biochemical markers, scanning using positron emission tomography or MRI, and endoscopic procedures such as endosonography and endoscopic retrograde cholangiopancreatography are discussed. A combination of techniques is often used in an attempt to diagnose early cholangiocarcinoma. Cholecystectomy should be performed for gallbladder polyps, as many are malignant, and ultrasonography and α-fetoprotein testing are suggested for screening for hepatocellular carcinoma. Colorectal carcinoma screening should be performed after the diagnosis of PSC, and surveillance colonoscopy should be performed annually if there is concomitant colitis.
- primary sclerosing cholangitis
- bile duct
Statistics from Altmetric.com
Primary sclerosing cholangitis (PSC) is a chronic inflammatory condition that causes fibrosis of the biliary tree. Its incidence is highest in young men, and approximately three-quarters of northern European patients with primary sclerosing cholangitis will have associated inflammatory bowel disease,1 most commonly ulcerative colitis (UC). Where PSC occurs with Crohn’s disease, the distribution of Crohn’s disease is often predominantly colonic. The median survival from diagnosis of PSC has traditionally been 10–12 years in symptomatic patients.2 3 This high mortality can be attributed to the high incidence of malignancy and liver failure. It is well recognised that PSC is associated with a high incidence of cholangiocarcinoma and increased risk of developing gallbladder cancer and possibly pancreatic cancer.1 Hepatocellular cancer (HCC) can occur in cirrhotic patients with PSC,4 and, in combination with inflammatory bowel disease (IBD), PSC increases the risk of colorectal cancers above the risk of colorectal cancer in IBD alone.5 Table 1 shows the increased incidence of these malignancies in patients with PSC compared with the general population.
In the light of this high prevalence of malignancy, this article will review the current evidence for chemoprevention and screening in PSC.
RISK FACTORS FOR MALIGNANCY IN PSC
Many studies have investigated the risk factors involved in the development of cancers in PSC. Knowledge of these factors may help us to target patients who are more likely to require chemoprevention or screening. Although smoking,6 alcohol,7 duration of IBD,8 and previous colorectal cancer/dysplasia9 have been implicated in individual studies to varying degrees as risk factors for cholangiocarcinoma, none have been replicated as being of use as a predictive factor. Moreover, the time since diagnosis of PSC and the severity of liver disease (Child–Pugh and Mayo Risk Score) have not been shown in multiple studies to be significant factors.7 10 In fact, a large proportion of cholangiocarcinomas (33–50% depending on the study) are diagnosed within 1 year of diagnosis of PSC.1 8 Subsequent development of cholangiocarcinoma occurs at a rate of 0.5–1.5% patients per year, with the cumulative lifetime risk of 10–15%.6 11 Some have suggested that the PSC-like changes found synchronously with cholangiocarcinoma at presentation may not be true PSC, but secondary to the cholangiocarcinoma. However, on review of the specimens and images of patients with a synchronous diagnosis of cholangiocarcinoma with PSC, and taking into consideration the distribution of the cholangiographic abnormalities relative to the localisation of the tumour and the history of pathological liver tests, Boberg et al8 were not convinced that the changes were secondary, and a co-existent diagnosis of PSC was kept in their patients. It is likely that PSC develops through a long asymptomatic phase when liver function tests may be normal, and cholangiocarcinomas that occur in this phase can cause symptoms that bring the patient to medical attention. These cholangiocarcinomas are then detected soon after and account for the initial high frequency of cholangiocarcinoma after presentation. This would explain the large proportion of cancers presenting within 1 year of diagnosis, and would skew the mean duration of PSC before the development of the cholangiocarcinoma.
Many studies have reported an increased risk of colorectal dysplasia/cancer in patients with IBD. Other risk factors that increase the risk of colorectal cancer in IBD, which may equally apply to patients with PSC, include the extent and duration of the disease12 and a family history of colorectal cancer.13 To a lesser degree, the severity of inflammation,14 as evidenced by pseudopolyps, and the lack of use of 5-aminosalicylates (5ASAs)15 have been found to be factors. The risk of colorectal cancer in a patient with pancolitis starts to increase after 8–10 years and increases progressively. The risk is 2% at 10 years, 5–10% at 20 years, and 12–20% at 30 years. In 1995, a study by Broome et al9 showed that the absolute cumulative risk of developing colorectal dysplasia or cancer in patients with PSC was 9%, 31% and 50%, respectively, after 10, 20 and 25 years duration of the disease compared with 2%, 5% and 10%, respectively, for patients with UC without PSC (fig 1).
Although most studies have shown an increased risk of colorectal cancer in patients with UC and PSC, some have not. In 2002, Soetikno et al5 performed a meta-analysis to clarify this issue, and found an increased risk of colorectal cancer in patients with UC and PSC compared with those with UC alone (odds ratio (OR) 4.79 (95% CI 3.58 to 6.41)). Figure 2 depicts the results of the meta-analysis showing the additional risk of developing colorectal neoplasia and carcinoma for PSC in patients with UC.5
Some groups have questioned whether the risk is increased by PSC itself or whether it is because the associated colitis is often pancolitis with a subclinical course.16 17 As a result, the colitis tends be diagnosed late and tends not to need colectomy for medically uncontrollable flare-ups, thereby increasing the risk of colorectal neoplasia, purely from the increased extent and duration of colitis.
It has been recognised that HCC can occur in cirrhotic livers from many causes, including PSC. The prevalence of HCC in patients with PSC undergoing liver transplantation was reported to be 2% in one study.4 This was thought to be comparable to the rate seen at transplantation for other cholestatic liver diseases, such as primary biliary cirrhosis.
There appears to be an increased risk of gallbladder cancer in patients with PSC who have gallbladder polyps. Buckles et al18 found that 57% of gallbladder polyps were malignant in patients with PSC. This is in contrast with gallbladder polyps in patients without PSC, where the polyps are generally thought to be benign if less than 10 mm in size. In addition, a recent study, by Lewis et al,19 looking at gallbladders in patients with end-stage PSC, found that 14% of the gallbladders harboured an adenocarcinoma. There was increased risk of gallbladder neoplasia in patients who had concomitant IBD or intrahepatic biliary neoplasia, but this was not related to the duration of PSC. The authors suggested that gallbladder dysplasia may be part of a “field effect” in biliary dysplasia in these patients.
No risk factors have been identified for the increased risk of pancreatic cancer, although it is interesting to note that pancreatic changes, both functionally and structurally, have been shown to occur in patients with PSC.20 21
An association has been found in a study by Broome et al9 between colonic and biliary dysplasia/malignancy in patients with PSC, and it has been postulated that there may be a genetic predisposition in these patients with PSC to the development of dysplasia/malignancy. Genetic studies have shown a trend towards increased risk of cholangiocarcinoma in patients with HLA− DR4, DQ8 haplotype.22 Another study found an increased association of specific matrix metalloproteinase alleles (MMP-1 allele 1G) in patients with PSC who have UC and cholangiocarcinomas.23 However, only 15 patients were studied in this subgroup, and further confirmation from other studies is required. No other specific genetic factors have been shown to be associated with the increase in malignant risk in patients with PSC.
Key points on risk factors for malignancy in PSC
There is an increased risk of cholangiocarcinoma, gallbladder cancer, hepatocellular carcinoma, colorectal cancer, and possibly pancreatic cancer in patients with primary sclerosing cholangitis (PSC).
Up to 50% of cholangiocarcinomas are detected within 1 year of diagnosis of PSC.
The lifetime risk of cholangiocarcinoma is ∼15%.
PSC is an additional risk factor for colorectal cancer in patients with ulcerative colitis.
Gallbladder polyps are more often malignant in patients with PSC.
With survival threatened by malignancy in this often young group of patients, the idea that these malignancies could be reduced with chemoprevention is an enticing one. Most of the evidence for the chemopreventive benefit of drugs has been found for those used for regular maintenance of PSC and the associated IBD. These include ursodeoxycholic acid (UDCA) and those in the 5ASA group.
UDCA is a hydrophilic bile salt found naturally in small amounts in human bile. Its exact mechanism in chemoprevention is unclear, but, in the liver, it is postulated to be protective by stabilising bile duct epithelium and hepatocyte cell membranes, displacing more toxic and hydrophobic bile salts from the bile pool24 and acting as a choleretic agent, increasing bile flow from the liver, thereby reducing intrahepatic bile stasis and exposure time to toxic bile salts.25
In vitro and animal studies have suggested that UDCA confers a chemoprotective effect in the colon as well. It has also been shown to reduce faecal concentrations of deoxycholic acid,24 which have been shown to be higher in patients with colorectal dysplasia and cancer, both those with26 and without27 UC. Other chemoprotective mechanisms have also been proposed, including disruption of changes in protein kinase C isoforms induced by carcinogens,28 changes in arachidonic acid metabolism in the colonic mucosa, and inhibition of cyclo-oxygenase 2 expression.29 The predominance of right-sided colonic cancers in PSC30 point to the possible effect of secondary carcinogenic bile acids, which, as mentioned above, may be influenced by UDCA use.
Two studies, by Tung et al31 and Pardi et al,32 have shown reduced rates of the development of dysplasia and colorectal cancer with the use of UDCA/ursodiol. Tung et al31 compared patients with UC who had PSC and were taking ursodiol with those not using ursodiol who were enrolled in their colonoscopic surveillance programme. They found a lower incidence of colonic dysplasia in those using ursodiol. Patients who were not using ursodiol had a significantly longer duration of colitis (21 vs 14.3 years), and the results remained significant after adjustment for this difference. When they looked at patients with high-grade dysplasia, they found a significantly reduced rate in those using ursodiol after adjustment for other variables (OR 0.16 (95% CI 0.03 to 0.96), p = 0.04). Pardi et al32 followed patients with PSC/UC who were enrolled in a previously randomised controlled trial in a colonoscopic surveillance programme, and compared those taking UDCA with those not taking UDCA. They found a reduced relative risk of dysplasia or cancer for those using UDCA compared with placebo (RR 0.26 (95% CI 0.06 to 0.92), p = 0.034) (fig 3).
A recent historical cohort study, aiming to confirm the beneficial effects of UDCA, by Wolf et al,33 comparing patients with PSC who were using UDCA with those who were not, found a non-significant reduction in the incidence of colorectal dysplasia or cancer, but a significant reduction in cumulative overall mortality in those using UDCA. However, there were no data on how patients were selected for UDCA or not and whether they received other drugs, including 5ASAs.
The benefit of UDCA on the rate of cholangiocarcinoma is harder to demonstrate, as 33–50% may occur within the first year after diagnosis of PSC. The likelihood of any preventive strategy working within this time is slim. In one retrospective study, Brandsaeter et al,34 looking at a PSC population referred for transplantation, found an increased risk of hepatobiliary cancers in patients who did not use UDCA. Olsson et al35 did not find a significant difference in the incidence of cholangiocarcinoma in a randomised controlled study comparing high-dose UDCA (17–23 mg/kg body weight/day) with placebo over 5 years. Although the mean duration of PSC disease in each group was 5–6 years, it was interesting to note that two of the three patients who developed cholangiocarcinoma in the UDCA arm did so within the first year of diagnosis of PSC. The authors also did not reach their target enrolment number and admit that the study may not be able to exclude a beneficial effect. Another recent study, by Rudolph et al,36 followed up patients with PSC who were taking UDCA. The mean duration of follow-up was 6.8 years. However, they report that the rate of cholangiocarcinoma in their cohort was lower than a historical control rate, and, in their study, no cholangiocarcinomas were detected in patients who had been using UDCA for longer than 8 years (n = 55). They suggest that their data support the assumption that UDCA reduces the rate of cholangiocarcinoma. Unfortunately, other published studies have not been able clarify this, as they have generally been smaller with shorter follow-up. However, anecdotally, other studies using UDCA, such as one by Stiehl et al,37 have reported low prevalence of cholangiocarcinoma. There is no evidence to evaluate the benefit of different doses of UDCA in chemoprevention. The results of the larger multicentre blinded and randomised trial sponsored by the National Institutes of Health on the effect of high-dose UDCA on clinical and survival end points will hopefully clarify the current position.38
5ASA and sulfasalazine are used in the treatment and maintenance of UC. They decrease the inflammation in the colonic mucosa. Proposed mechanisms of 5ASAs in decreasing the risk of colorectal cancers include induction of apoptosis, decreasing cell proliferation, and inhibition of nuclear factor κB.39
Many case–control studies40 and a recent meta-analysis15 have shown 5ASA to be of benefit in the prevention of colorectal cancer, with an increasing effect the longer the duration of use. The minimal dose for an effect appears to be 1.2 g/day. However, specific studies of the effect of 5ASAs in patients with UC complicated by PSC are still lacking. Lindberg et al30 investigated the effect of sulfasalazine in this group, and found a non-significant difference in the rate of colorectal dysplasia or cancer (44% vs 34% in those not using and using sulfasalazine, respectively). The authors suggested that a larger sample size may have yielded significant results. There are two further possible explanations for why no benefit was found with sulfasalazine in this population. Firstly, sulfasalazine may be less effective than other 5ASAs in chemoprevention. Unlike other 5ASAs, it is a competitive inhibitor of folic acid absorption, and this may negate the beneficial effects seen with other 5ASAs. Two studies41 42 have shown that sulfasalazine is associated with a lower colorectal cancer risk reduction than other 5ASAs, although this may be due to the longer duration of IBD in patients using sulfasalazine than in those using other 5ASAs.
Secondly, there may be differences in the colon in patients with UC with PSC and those without that may affect the efficacy of drugs in the 5ASA group in preventing dysplasia. There is evidence that the severity of inflammation in the colon is a risk factor in the development of colorectal cancer in patients with UC.14 Previous studies have reported a predisposition to subclinical manifestation and a milder clinical course of the associated colitis in patients with PSC.16 5ASA may therefore be of less use in reducing active inflammation in the colon of patients with PSC. In addition, Heuschen et al43 have suggested that toxic bile acids have an influence in the aetiology of proximal colorectal cancers and associated back-wash ileitis in PSC. For these reasons, the benefit of 5ASA as a chemoprotective agent may be decreased in patients with PSC. Further studies on patients with PSC are required, but randomised controlled trials are unlikely as it would be unethical to withhold 5ASAs, as they are routinely used in the treatment and maintenance of UC.
There is no evidence for the benefit of other drugs used for maintenance in colitis/IBD, such as azathioprine and 6-mercaptopurine.44
Chemoprevention has not been studied in other cancers associated with PSC.
Key points on chemoprevention
Ursodeoxycholic acid (UDCA) appears to be chemoprotective against colorectal neoplasia in patients with primary sclerosing cholangitis (PSC).
There is some suggestion that UDCA may also be chemoprotective for cholangiocarcinoma.
The role of 5-aminosalicylates in chemoprevention of colorectal cancer in patients with PSC and ulcerative colitis is still unclear.
The aim of screening in PSC is to detect cancers at an early stage and improve the prognosis of these patients. As in any screening process, this is best done when the natural history is well known, a pre-malignant stage may be identified with good sensitivity and specificity, and an effective treatment is available that will improve the final outcome of the patient. The sequence of dysplasia–neoplasia in the colon in patients with IBD is accepted, and there is increasing evidence for this sequential progression in the biliary tract and gallbladder in patients with PSC (discussed below). High-grade dysplasia in the colon (fig 4) is associated with synchronous and geographically separate malignancy, and its presence is an indication for colectomy. However, despite reports of biliary dysplasia being associated with or occurring before cholangiocarcinoma, its presence in a liver biopsy specimen has not universally been accepted as an indication for transplantation.
The natural history and carcinogenesis pathway for cholangiocarcinoma is not clear, but there is some evidence that biliary dysplasia is a precursor of cholangiocarcinoma. Case–control studies45 46 have suggested that biliary dysplasia is found more often on liver biopsy of patients with PSC with cholangiocarcinoma than those without. Moreover, the dysplastic areas can also be found in liver tissue geographically separate from the cholangiocarcinoma.6 Other studies have also suggested that the presence of cytological abnormalities, from bile duct brushings, correlating with biliary dysplasia, are more common in patients with cholangiocarcinoma.47
However, detection of early cholangiocarcinoma and its differentiation from benign strictures seen in PSC is difficult. Many patients are found with cholangiocarcinoma at transplantation despite intensive work-up before transplantation,3 6 34 48 indicating the difficulty in diagnosing these cancers. Screening regimens and techniques to diagnose early cholangiocarcinomas are discussed below.
The possible treatment options for cholangiocarcinoma include surgical resection and transplantation. Resection for cholangiocarcinoma on a background of PSC is difficult because there may be underlying fibrosis and cirrhosis, there may be dysplastic field change with synchronous lesions or recurrence after resection, and reported long-term survival after resection is poor, with <10% survival at 5 years. As a result, some units do not perform resections on these patients but refer them for transplantation.49
The outcome of transplantation for PSC without malignancy is very good,50 51 and the European Liver Transplant Registry and the Nordic countries52 have reported outcomes comparable to transplantation for other indications. The 5 and 10 year survival of patients with PSC from the European Liver Transplant Registry (1988–2006) are 76% and 66%, respectively. This is comparable to the survival rate during the same period for primary biliary cirrhosis (78% and 69% for 5 and 10 year survival, respectively) and alcoholic cirrhosis (72% and 56%, respectively). Therefore it has been suggested that patients with PSC should have a transplantation before any cancers occur. However, as no definitive risk factor or scoring system has been identified to predict cholangiocarcinoma in PSC, it is unreasonable to expect every patient with PSC to undergo a pre-emptive transplantation, and optimal timing remains difficult. Different groups have suggested different clinical and biochemical criteria for early transplantation,53 54 and others have tried to predict future neoplastic change with biliary brush cytology, and have referred these patients for liver transplantation.55 Historically, cholangiocarcinoma has been regarded unfavourably for transplantation because of early recurrence and poor survival after transplantation.54 However, at least one group, Goss et al,56 has found no worsening of survival in those that were found to have incidental cholangiocarcinoma in their explanted liver. This, however, has not been consistent for all groups.34 54 Brandsaeter et al34 reported an overall 5-year survival of 35% for patients found to have cholangiocarcinoma when transplanted with a suspected hepatobiliary malignancy but no identifiable lesion.
In addition, a few transplant units have transplanted highly selected patients with unresectable hilar cholangiocarcinoma after neoadjuvant radiochemotherapy with good long-term results.49 57 58 Although this occurs in only a few selected units at the moment, there is hope that there may be wider availability of these programmes in the future.
Investigations for cholangiocarcinoma are often initiated after symptoms or new biochemical changes with the use of tumour markers (such as CA19-9), non-invasive imaging (such as ultrasonography, CT or MRI/magnetic resonance cholangiopancreatography (MRCP)), and invasive procedures (such as endoscopic retrograde cholangiopancreatography (ERCP)), often with brushings. However, the diagnosis for cholangiocarcinoma is often difficult despite these tests, and the cancer is often metastatic by the time of diagnosis.8 A median survival of 5 months has been reported after diagnosis of cholangiocarcinoma in PSC.59
Although there are no data on the benefit of screening for cholangiocarcinoma at present, for the reasons above, many groups are investigating methods of detecting the disease at an early or even pre-malignant stage. Tumour markers, although appealing as a screening tool for the clinician, are unreliable in the early detection of cholangiocarcinoma, are raised in other non-malignant diseases,60 and often rise late in the course of the cancer. The early promise of the accuracy of tumour markers, either alone or in combination, such as the index of two serum tumour markers (CA19-9 + (CEA×40) >400) as suggested by Ramage et al,61 has not been confirmed in later studies.7 62
The combination of tumour markers with ERCP has also been investigated—for example, by Siquiera et al48—but, unfortunately, no prospective validating studies have been performed to date to show any increased benefit in their combined use in predicting the occurrence of early cholangiocarcinoma.
MRI/MRCP has traditionally been the initial imaging investigation of choice in the work-up of cholangiocarcinoma,60 but it is still often difficult to differentiate between benign strictures and early malignant strictures.63 64 The general improvement in imaging technology has meant that other imaging modalities may have a role in the detection and differentiation of early cholangiocarcinomas. Multislice CT cholangiography has recently been reported to have promise in the imaging of the obstructed biliary tree in comparison with MRCP.65 However, it has not been tested in PSC, and its utility in this setting remains to be proven.
Another modality under investigation is positron emission tomography (PET) scanning. This method detects positron-emitting radiolabelled tracers, in this case [18F]fluoro-2-deoxy-d-glucose (FDG), a glucose analogue that is phosphorylated in the cells but not further metabolised. Tumour cells accumulate FDG because of their high metabolic rate and can theoretically be differentiated from normal tissue. However, there has been disagreement about the benefits of FDG PET scanning in differentiating cholangiocarcinoma from benign strictures in patients with PSC.66 67 A recent study of the use of dynamic, rather than static, FDG PET scanning in a pre-transplant setting seemed to show promise in detecting early cholangiocarcinoma, but was unable to detect biliary dysplasia.68 However, its benefit will need to be confirmed by other studies, and therefore its role, perhaps in determining which patients should undergo intra-arterial chemotherapy before transplantation, remains to be clarified.
Although endoscopic ultrasound has been shown to be useful in the management of cholangiocarcinoma, its use in detecting early cholangiocarcinoma specifically in patients with PSC has not been fully investigated. Bile duct imaging criteria, reported by Lee et al,69 suggestive of malignancy, including irregular bile duct wall and a thickness of more than 3 mm, may occur in benign strictures of PSC, and difficulties have been found in differentiating between the two.70 71 Endoscopic cholangioscopy has not gained widespread use, mainly for technical reasons, but there is hope that future probes may allow easier directed tissue sampling.72 A recent study, Tischendorf et al,73 has looked at the use of cholangioscopy in patients with PSC and dominant bile duct strictures and found high sensitivity and specificity (92% and 93%, respectively) in detecting cholangiocarcinoma in these strictures. In contrast, Awadallah et al,74 who also looked at the use of cholangioscopy and biopsies in patients with a dominant stricture, failed to detect two out of three cholangiocarcinomas in 40 patients. The authors reported that these patients had cholangioscopy through pre-existing percutaneous tracts, which they thought may have limited their examinations. However, no studies have been performed using cholangioscopy as a screening test for cholangiocarcinoma in patients with PSC.
The use of ERCP with bile duct brushings is well established in the investigation of cholangiocarcinoma in patients with PSC. Most cholangiocarcinomas that develop on a background of PSC are extrahepatic and accessible by this route. However, most studies show a sensitivity of 17–60% in the detection of cholangiocarcinoma for routine cytology.75–77 Interpretation may be made more difficult by the reactive and inflammatory changes in the bile duct caused by PSC. Some additional procedures may improve the sensitivity and specificity of brushings, including dilatation of the stricture before brushings, repeated brushings,48 78 and the use of additional advanced analysis of cytological specimens, such as digital image analysis (DIA) and fluorescent in-situ hybridisation (FISH). DIA measures cellular DNA content and quantifies aneuploidy in the cells. FISH uses commercially available fluorescent DNA probes, which hybridise to centromeres of chromosomes 3, 7, and 17 and the p16 gene on chromosome 9. After a counter stain, fluorescent microscopy can be used to detect abnormal cells. Both were found to increase sensitivity above routine cytology. Moreno Luna et al76 showed that both DIA and FISH increased sensitivity by an additional 14% and 60%, respectively, on top of routine cytology (sensitivity 17%) in patients with PSC and strictures. When DIA and FISH were combined, the specificity was very high (98%), but at a cost of decreased sensitivity (but still an additional 14% above routine cytology).
Despite these tests, there are very few data on screening asymptomatic PSC patients for pre-malignant changes or early cholangiocarcinoma. This is understandable given the possible complications of ERCP, including cholangitis and pancreatitis as well as the limitations of curative treatment for cholangiocarcinoma in this setting. In addition, no cost–benefit analysis has been performed on the use of any screening procedures for cholangiocarcinoma.
However, two studies are of some interest. Boberg et al,47 performed a prospective study looking at the benefit of bile duct brushings of strictures in 61 consecutive patients with PSC, and comparisons were made with findings by liver histology, where available. A median follow-up of 38 months or post-transplant outcomes were described. They reported a sensitivity, specificity and positive and negative predictive value for biliary dysplasia/malignancy of 100%, 84%, 68% and 100%, with an overall accuracy of 88%. Sensitivity decreased and specificity increased when biliary cytology with only high-grade dysplasia was used. Those that were found to have cholangiocarcinoma had poor survival, but seven patients with biliary dysplasia underwent liver transplantation as a result of the cytological findings and appear to have no evidence of recurrence after a median follow-up of 27 months. The accuracy in detecting biliary malignancy in this study is higher than reported, and questions have been raised about the generalisation of their results to other less specialised institutions. The results of the study suggest a possible benefit in pursuing biliary cytology to detect early changes/dysplasia in improving the long-term outcome of patients with PSC. However, ERCP was performed for “generally accepted” indications in the study, and the poor outcomes of those found to have cholangiocarcinoma suggest that this may be too late in many cases.
Moff et al55 tried to address this question by reviewing data from patients undergoing ERCPs as part of a screening and surveillance programme. These ERCPs were performed in patients without any abnormal changes in imaging or biochemical markers. Despite this proposed use of ERCP as a screening tool, the mean bilirubin concentration was raised (2.5–3.0 mg/dl), so the patient population in the study may have been at a relatively advanced stage of PSC. In the study, bile duct brushings were obtained in the right, left and common hepatic ducts as well as the common bile ducts from 47 patients, even when no strictures were seen. Of the three patients who had had a liver transplant, with neoadjuvant radiochemotherapy, for marked atypia on their brushings in their unit, two were found to have cholangiocarcinoma in their explants, with no reported recurrence after transplantation. Three others with marked atypia were undergoing assessment for transplantation at the time of publication. Two out of 31 patients in whom brushings were thought to be benign developed peripheral cholangiocarcinomas detected on cross-sectional imaging. The other 29 patients with benign brushings have not developed cholangiocarcinomas during follow-up. Complications were limited to mild pancreatitis (n = 1), bleeding (one mild and one moderate) and cholangitis due to stent blockages (n = 2) in 101 ERCP procedures. Although this study is of great interest in suggesting the possibility of screening ERCPs in patients with PSC, many questions remain unanswered. No details are given about when or which patients should be entered into a screening programme, or the interval between each ERCP used in the study. Future prospective studies will be required to assess the cost–benefit of screening using combinations of the above tests, as well as clarifying the optimum method, timing and patient selection in any future programmes.
It remains likely that current “screening” for cholangiocarcinoma will be a combination of different modalities. Several groups have recommended yearly MRI/MRCP scans, with ERCP and cytological examination (including FISH and DIA, where available) in those that develop strictures or changes in biochemical variables or tumour markers. The benefit of using these techniques and their combinations for screening will need to be assessed prospectively, but the impetus for screening, particularly in asymptomatic patients, will ultimately depend on the available treatment options and their outcome. The preferred balance between sensitivity and specificity of the tests in screening for cholangiocarcinoma will be determined by the changes in indications and outcomes for transplantation for patients with PSC and cholangiocarcinoma. In addition, further investigation into genetic studies is needed to find patients at increased risk of cholangiocarcinoma who might benefit from an early or “pre-emptive” transplantation.
Key points on screening for cholangiocarcinoma
There is no accepted screening programme for cholangiocarcinoma in asymptomatic patients with primary sclerosing cholangitis.
Curative options for confirmed cholangiocarcinoma in these patients are limited.
Early cholangiocarcinomas are difficult to differentiate from benign strictures.
Tumour markers are not sensitive enough to screen for early cholangiocarcinomas.
Biliary dysplasia should be viewed as a pre-malignant condition.
Standard investigations for cholangiocarcinoma with MRI/MRCP and ERCP with brush cytology still miss a significant proportion of cholangiocarcinoma
Other modalities, such as dynamic FDG-PET, FISH, DIA and cholangioscopy, may be useful to in detecting early cholangiocarcinomas.
No data exist for screening programmes for gallbladder cancer. Lewis et al19 proposed that gallbladder cancers in PSC follow the metaplasia–dysplasia–carcinoma sequence, and there have been suggestions that routine, repeated ultrasound scans should be performed on patients with PSC, even in stable disease, to detect gallbladder polyps. Cholecystectomy should be recommended if polyps are detected in the gallbladder, as these are often malignant or can progress to gallbladder carcinoma.18 79 The interval and cost-effectiveness of this strategy has not been determined.
Guidelines for HCC80 recommend screening once the patient is thought to be cirrhotic, with α-fetoprotein testing and imaging with either ultrasonography or MRI if additional assessment of prevalent strictures is required. There are no randomised controlled trials in proving the benefit of screening. The recommended screening interval is 6-monthly, although patients with a non-diagnostic but rising α-fetoprotein concentration have been recommended to have 3-monthly ultrasonography, in view of the likely increased risk. Outcomes from transplantation of patients with HCC within the Milan criteria (single lesion of less than 5 cm in diameter or three lesions or less with a maximal diameter of less than 3 cm) show a prognosis equal to transplantation for the underlying liver disease without HCC,80 and Brandsaeter et al34 report promising long-term outcomes for transplantation of patients with PSC and HCC less than 5 cm in the Nordic countries.
There is no screening programme for pancreatic cancer, and some have questioned whether the reported increased risk in pancreatic cancer is real, as small distal cholangiocarcinomas are often difficult to differentiate from pancreatic cancer and the findings of increased pancreatic cancer in patients with PSC have not been replicated in other studies.81
Screening for colorectal dysplasia and cancer in patients with PSC has generally been instituted and accepted, despite no clear randomised controlled trials, in those with coexistent UC. Guidelines are available from many gastroenterological societies,82 83 with recommendations for surveillance from the time of diagnosis of colitis and for shorter intervals, such as annually, between colonoscopies in patients with PSC because colitis in these patients is often subclinical and therefore often of unknown duration. Patients with PSC should undergo a colonoscopy at the time of diagnosis of PSC to look for the development of colitis even if they are asymptomatic.16 At our institution, a repeat colonoscopy is recommended after 4 years despite a normal initial colonoscopy and biopsies, as colitis can occur after the diagnosis of PSC. Other units only repeat colonoscopy when there are new symptoms suggestive of colitis. There are no published data on the risks of colorectal cancer in patients with Crohn’s disease with coexistent PSC, but, as increased risk of colorectal neoplasia has been reported in those with longstanding Crohn’s colitis,84 recommendations have been made that surveillance should similarly apply to those with Crohn’s colitis. Our own as yet unpublished data suggest that the risk of dysplasia in patients with Crohn’s disease and PSC is not as high as in UC and PSC. However, units may vary in their practice of screening of these patients.
In patients with PSC and UC, it is important to examine the colon completely, with particular attention paid to the right colon. There appears to be increased risk of a right-sided distribution of the colonic neoplasia in patients with PSC (up to 75% of all found in one study30) compared with patients with UC alone.
Most dysplastic areas (61–77%) are thought to be detectable by normal, thorough, colonoscopy,85 although training in recognising flat dysplastic lesions has been recommended.86 Traditionally, biopsies have been performed randomly quadrantically every 10 cm with additional targeted biopsies of abnormal areas and their surrounding mucosa.82 Particular attention should be paid to the right colon in patients with PSC 30 and for dysplasia-associated lesions or masses and flat dysplastic lesions. Although it has been suggested that it is safe to treat dysplasia found in adenoma-like lesions with endoscopic removal,87 88 detection of high-grade dysplasia, field change or dysplasia-associated lesions or masses, in addition to carcinoma, warrants colectomy.89 90 Treatment of low-grade dysplasia has been more controversial, partly because there is a high level of inter-observer variation, and partly because various studies have not agreed about its degree of correlation with malignancy or progression. However, a recent meta-analysis found an overall positive predictive value of low-grade dysplasia for concurrent colorectal cancer to be 25.5%.91 The treatment recommendations vary from shorter intervals of surveillance to colectomy, particularly if repeatedly found on consecutive colonoscopies and confirmed by two independent gastrointestinal pathologists.91 92
Some reports have questioned the use of surveillance programmes, and surveys have found particular areas of practice to be deficient. Eaden et al93 found that most endoscopists did not comply with the recommended number of biopsies, and only 53% recommended colectomy when high-grade dysplasia was found in the specimens. Patient surveys94 have highlighted that a proportion of patients were unaware of the need for surgery should carcinoma be found on screening. Lynch et al95 have questioned the overall benefit of the surveillance programme in detecting and preventing deaths from colorectal cancer. However, other studies96 97 have suggested that, for patients who are in surveillance programmes, the diagnosis of colorectal cancer is made at an earlier stage, resulting in a potentially better long-term prognosis, than for those who are not.
Bergquist A, Ekbom A, Olsson R, et al. Hepatic and extrahepatic malignancies in primary sclerosing cholangitis. J Hepatol 2002;36:321–.7
Goss JA, Shackleton CR, Farmer DG, et al. Orthotopic liver transplantation for primary sclerosing cholangitis. Ann Surg 1997;225:472–83.
Heimbach JK, Haddock MG, Alberts SR, et al. Transplantation for hilar cholangiocarcinoma. Liver Transpl 2004;10(10 Suppl 2):S65–8.
Fleming KA, Boberg KM, Glaumann H, et al. Biliary dysplasia as a marker of cholangiocarcinoma in primary sclerosing cholangitis. J Hepatol 2001;34:360–5.
Soetikno RM, Lin OS, Heidenreich PA, Young HS, Blackstone MO. Increased risk of colorectal neoplasia in patients with primary sclerosing cholangitis and UC: a meta-analysis. Gastrointest Endosc 2002;56:48–54.
Keisslich R, Neurath MF. Surveillance colonoscopy in UC: magnifying chromoendoscopy in the spotlight. Gut 2004;53:165–7.
Further evidence suggests a benefit of using targeted biopsies in combination with chromoendoscopy. Studies have compared the use of indigo carmine98 and methylene blue99 with the recommended quadrantric random biopsies. The results have shown higher pick up rates for dysplasia. In a randomised trial using methylene blue, Kiesslich et al99 showed that there was about a threefold increased detection of intraepithelial neoplasia, particularly flat lesions, when using chromoendoscopy (table 2).
New guidelines for the use of these techniques have been suggested.100 These guidelines have been called SURFACE (strict patient selection; unmask the mucosal surface; reduce peristaltic waves; full-length staining of the colon; augmented detection with dyes; crypt architecture analysis; endoscopic targeted biopsies). Other newer imaging techniques, such as narrow band imaging101 and confocal laser endomicroscopy,102 are currently being assessed but have not filtered down to standard clinical practice.
Figure 5 shows the investigations used in our institution for screening and surveillance in patients with asymptomatic PSC with unremarkable biochemistry. Patients undergo colonoscopy at diagnosis of PSC if they have not already had a diagnosis of IBD, and subsequently have yearly or 4-yearly colonoscopies depending on whether colitis is found or not, respectively. Table 3 gives a summary of suggested screening and surveillance tests for cancer in patients with PSC.
Key points on screening for cancers associated with primary sclerosing cholangitis (PSC)
No studies have been performed to assess the benefit of screening for gallbladder cancer, hepatocellular cancer and pancreatic cancer in patients with PSC.
Ultrasonography has been suggested for gallbladder polyps and lesions, which should be removed by cholecystectomy on detection.
6-monthly ultrasonography and α-fetoprotein testing have been suggested for hepatocellular carcinoma in patients with cirrhosis.
Colonoscopy should be performed, even in the absence of colonic symptoms, in patients with PSC to exclude colitis.
Yearly colonoscopic surveillance should be offered to patients with PSC and ulcerative colitis.
Chromoendoscopy improves the detection rate of dysplastic lesions in the colon.
PSC is a chronic cholestatic liver disease which increases the risk of malignancy. Survival in patients with PSC is decreased from both liver failure and malignancy. The two most common cancers are colorectal cancer and cholangiocarcinoma. There appears to be a protective benefit in using 5ASAs in patients with UC, but whether this also applies to those with concomitant PSC has not yet been confirmed. Some studies have suggested a benefit of UDCA as a chemopreventive agent in the colon and perhaps also in protecting against cholangiocarcinoma.
Colorectal cancer screening in patients with UC is well established, and, in those with PSC, the screening interval is shortened to annual testing irrespective of the duration of colitis. It has been suggested that this should also apply to patients with PSC and Crohn’s colitis. An initial colonoscopy should be performed in patients with PSC without bowel symptoms to diagnose mild or subclinical UC, and patients in screening programmes appear to have an improved survival. Recent evidence suggests that colonoscopy should be performed with dye spray and targeted biopsy to improve detection of dysplasia.
Routine ultrasound has been suggested for the screening of gallbladder cancers, and cholecystectomy is recommended when gallbladder polyps are found. There are no recommendations for formal screening programmes for cholangiocarcinoma at present, although there is continued interest in improving early detection of pre-malignant and malignant lesions. Treatment options for cholangiocarcinoma remain limited. However, newer methods and better outcomes for transplantation in selected cases may change future practice.
SELF ASSESSMENT QUESTIONS (TRUE (T)/FALSE (F); ANSWERS AFTER THE REFERENCES)
The risk of cholangiocarcinoma correlates with the duration of PSC.
Patients with PSC and UC have a higher risk of colorectal neoplasia than those with UC alone.
In patients with PSC and UC, the colorectal cancers tend to be more proximal than in those with UC alone.
Tumour markers, such as CA19-9, are useful in screening for early cholangiocarcinoma.
Current guidelines suggest that MRI is the best initial investigation for suspected cholangiocarcinoma.
Incidental gallbladder polyps in patients with PSC should be removed with cholecystectomy.
Transplantation is currently the standard treatment for cholangiocarcinoma.
Detection of high-grade dysplasia in the colon is an indication for colectomy.
UDCA is beneficial in reducing colonic dysplasia in patients with PSC.
The lifetime risk of cholangiocarcinoma in patients with PSC is 0.5–2%.
1. F; 2. T; 3. T; 4. F; 5. T; 6. T; 7. F; 8. T; 9. T; 10. F
Competing interests: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.