|Home | About | Journals | Submit | Contact Us | Français|
Therapeutic strategies for autoimmune liver diseases are increasingly established. Although proportionately uncommon, specialist centers have with time refined the best approaches for each disease, based on an improved understanding of the spectrum of presentation. The major treatment aims are to prevent end-stage liver disease and its associated complications. As a result of drugs such as ursodeoxycholic acid, predniso(lo)ne and azathioprine, both primary biliary cirrhosis and autoimmune hepatitis are now less commonly indications for liver transplantation. Unfortunately, the same inroads in treatment efficacy have as yet not been made for primary sclerosing cholangitis, although the recognition that a subset of patients may have a treatable secondary sclerosing cholangitis (IgG4 related) is helping a proportion. With better biological understanding, more specific interventions are expected that will benefit all those with autoimmune liver diseases.
Autoimmune liver diseases are chronic inflammatory processes in which an immunological attack is directed against hepatocytes and/or the biliary epithelium [Kumagi et al. 2008]. Table 1 illustrates the presently recognized common disease divisions. Such divisions may be artificial since broadly it is suggested that 10% of those with primary biliary cirrhosis (PBC) will have features of autoimmune hepatitis (AIH) [Chazouilleres et al. 1998; Czaja, 1998], 10% of adults with AIH have radiological evidence for sclerosing cholangitis (SC) [Abdalian et al. 2008] and a similar number of those with primary sclerosing cholangitis (PSC) will have histological evidence of AIH [van Buuren et al. 2000]. The rate of overlap seems higher in children with AIH, in whom 50% will have sclerosing cholangitis on endoscopic retrograde cholangio-pancreatography (ERCP) [Gregorio et al. 2001].
The diseases reflect a balance of environmental and genetic factors [Weismuller et al. 2008; Mackay, 2007] and whilst broadly autoimmune in etiology, both immune and nonimmune pathways are likely involved in tissue damage. Once the balance is tipped towards tissue destruction a chronic process ensues and clinicians are now well placed to modify disease by direct immunosuppression (e.g. predniso(lo)ne/ azathioprine) or indirect modification to the bile acid milieu (e.g. ursodeoxycholic acid [UDCA]). These diseases are uncommon, may progress slowly, and are recognized now to have a broad spectrum of presentation. Together these make clinical trial design and interpretation harder but herein we review the basis for the use of disease-modifying agents. Symptom control is covered more fully elsewhere [Hirschfield et al. 2008] and similarly the management of the poorly defined autoimmune overlap syndromes is also not detailed [Rust and Beuers, 2008].
AIH is a chronic and relapsing necroinflamma-tory hepatitis accompanied by elevated immuno-globulins and autoantibodies that help serological classification [Krawitt, 2006; Manns and Vogel, 2006; Czaja and Freese, 2002]. Type 1 disease is associated with nonspecific antinuclear and/or antismooth muscle antibodies, occurs across all ages, in women more so than men, and associates in Caucasians with HLA-DR3 (DRB10301) and DR4 (DRB10401). Patients with the DRB10301 allele tend to be younger, with more severe relapsing disease and have a greater chance of treatment failure. Type 2 AIH, representing less than 10% of all AIH, is defined by the presence of specific liver–kidney microsomal antibodies, occurs more frequently in Europe and South America, and nearly always presents in childhood, often with more severe disease. As a diagnosis of exclusion the mimics of disease are important to consider before starting immuno-suppression. Viral hepatitis must be ruled out but the other notable conditions include Wilson disease [Milkiewicz et al. 2000], alpha-1-antitrypsin deficiency [Lok et al. 1994] and drug-induced hepatitis [Liu and Kaplowitz, 2002]. Drug injury can replicate all features of AIH (Table 2) and stopping the offending medication usually, but not always, suffices in management.
There is a wide spectrum of presentation from acute liver failure (rare; 510% of all acute liver failure), all the way through to asymptomatic transaminitis (perhaps 30–35% of cases) [Werner et al. 2008; Feld et al. 2005; Kogan et al. 2002]. One-third of patients can expect to be cirrhotic at presentation, yet decompensated disease can favourably respond to treatment, assuming active inflammation remains. It is now recognized that one in five adults who develop disease are over 60: treatment, the rate of remission and relapse being similar [Al-Chalabi et al. 2006; Czaja and Carpenter, 2006]. Importantly the clinical presentation, disease behavior and treatment vary between racial groups. Black Americans, especially men, can have aggressive disease at presentation, are less likely to respond to conventional immunosuppressive therapy and have a higher mortality. As well they are younger and more likely to have a cholestatic variant of disease, which is not steroid responsive [Verma et al. 2007; D'Souza et al. 2005; Zolfino et al. 2002; Lim et al. 2001].
A number of scoring systems may help arrive at a diagnosis, although historically these originated for research, and are not prognostic [Alvarez et al. 1999]. The latest collaborative scoring system is appealing as it focuses on just autoanti-bodies, immunoglobulins, histology and exclusion of viral hepatitis [Hennes et al. 2008]. Liver biopsy is essential (assuming an appropriately sized biopsy), as biochemical markers need not correlate with histological severity. Despite this the classical features of a plasma cell infiltrate invading the limiting plate are not specific.
The major trials evaluating immunosuppressive therapy in AIH are relatively old, and focus on treating advanced disease. Since clinicians face treatment conundrums for patients with less advanced disease this does result in gaps in our knowledge. The studies nevertheless critically demonstrated that prednisone alone or in combination with azathioprine improved symptoms, laboratory tests, histologic findings and survival [Murray-Lyon et al. 1973; Soloway et al. 1972; Cook et al. 1971]. Importantly the trials do not demonstrate azathioprine monotherapy to be effective in inducing remission; its role is in maintenance therapy. Once remission has been obtained, the introduction of azathioprine permits the dose of corticosteroids to be reduced [Summerskill et al. 1975]. In one trial, azathiopr-ine (75 mg/day) was stopped and prednisolone continued at a maintenance dose of 5–12.5 mg/ day. The probability of relapse within three years was 32% in patients who stopped azathioprine compared with 6% in those who continued combination therapy [Stellon et al. 1985]. In a further study, patients were randomized to high dose azathioprine (2 mg/kg/day) after withdrawal of prednisolone or continued combination therapy (prednisolone 5–10mg/day and azathio-prine 1 mg/kg/day). There were no significant differences between the two groups and no patient relapsed on azathioprine alone at one-year follow up [Stellon et al. 1988]. Biochemical and histolo-gical remission was subsequently sustained in 83% of patients on azathioprine over the longer term [Johnson et al. 1995].
The absolute indications for treatment are based on the original studies of symptomatic patients and remain (1) serum aspartate aminotransferase (AST) greater than ten times upper normal, or (2) an AST greater than five-fold elevated with a greater than two-fold elevation in IgG, or (3) histological evidence of bridging or multiacinar necrosis. These criteria represent a severe end of the disease spectrum, in which untreated patients have a mortality of 40% within six months and ten-year survival of only 27% [Czaja and Freese, 2002; Soloway et al. 1972]. The absence of fibrosis or cirrhosis is not a relevant guiding factor. Patients outside such criteria are clearly still appropriate to treat but individual judgment is exercised, and a composite assessment made of symptoms, liver biochemistry and histology. The older literature suggests those with mild interface hepatitis have a normal five-year survival and a five-year progression to cirrhosis of around 20%, whilst mild-to-moderate laboratory abnormalities are associated with cirrhosis in 49% within 15 years but a ten-year survival of 90% [De Groote et al. 1978; Schalm et al. 1977; Baggenstoss et al. 1972]. Just as the magnitude of biochemical changes do not correlate with histological injury, nor is there any distinguishing feature between those with and without symptoms [Kogan et al. 2002]. The hope of treating more mild disease is to prevent progression of disease and cirrhosis because corticosteroids may prevent and reverse fibrosis [Czaja and Carpenter, 2004; Dufour et al. 1997] although this may not necessarily always be borne out [Roberts et al. 1996]. Not treating may also be valid: retrospectively in our practice, asymptomatic patients had lower serum aminotransfer-ase, total bilirubin and IgG at baseline and half received no therapy and their survival was no different to those with symptoms [Feld et al. 2005]. Patients with an inactive cirrhosis are not candidates for medical therapy and in those progressing rapidly to acute liver failure transplantation is usually more appropriate than steroids [Ichai et al. 2007].
Two general approaches evolved: (1) prednisone monotherapy; and (2) combination therapy, either from the outset, or with addition of azathioprine (or 6-mercaptopurine if preferred) later (Table 3). Both strategies work but most clinicians now use a combination approach, since this is associated with a lower occurrence of corticosteroid-related side effects (10% vs 44%) [Summerskill et al. 1975]. These two agents clearly have side effects (Table 4) and the monitoring of patients is well established [Czaja, 2008]. UDCA has been used for patients with mild disease in Japan, but this is not widely practiced elsewhere [Nakamura et al. 1998].
Response to immunosuppression therapy is assessed clinically, biochemically and by histology. In general a biochemical response (a decrease in serum aminotransferase and globulins), occurs within one to three months. A complete biochemical response is the aim, and a failure to see this should always lead to reappraisal (assuming compliance). Remission is considered normalization of transaminases, IgG, and histological activity and around two-thirds of patients will enter remission within 18 months and nearly 90% will achieve this after 3 years of treatment [Czaja et al. 1984, 1983; Soloway et al. 1972]. Histological remission lags behind clinical and biochemical remission by three to six months explaining prolonged maintenance therapy and the use of histology by some as the final endpoint. In general, patients with mild or asymptomatic disease have better responses, although patients with established cirrhosis at presentation, can achieve remission successfully with a ten-year life expectancy of at least 90% [Roberts et al. 1996]; almost all require life-long therapy as treatment withdrawal invariably leads to relapse in this context.
With remission, tapering of immunosuppression is appropriate; for example, if by two to three months serum aminotransferase levels are normal, the prednisone dose can be decreased by 2.5–5mg every two to four weeks. The precise strategy is center- and patient-specific, and in part relates to the initial dose of corticosteroid chosen. One typical approach [Johnson et al. 1995] has been the achievement of complete clinical, biochemical and histologic remission with prednisolone (5–15 mg/day) and azathiopr-ine (1 mg/kg/day; traditionally not initiated until serum bilirubin5100 mmol/l) for at least one year before the prednisolone is withdrawn. Long-term remission is maintained by low-dose corticoster-oid alone [Cook et al. 1971], or in combination with azathioprine (1 mg/kg/day) [Stellon et al. 1985]. Maintenance with azathioprine mono-therapy is well established and once complete remission has been induced and sustained the majority of patients will remain in remission with azathioprine at a slightly higher dose (2 mg/kg) given alone [Johnson et al. 1995]. After a further period of remission it is possible to maintain remission at a dose of 1mg/kg/day, and indeed many simply use 1 mg/kg from the outset. Some patients may need low-dose prednisone with the azathioprine or the higher azathioprine dose to maintain remission.
Although a chronic relapsing disease, there is a natural desire for discontinuing treatment and there is no agreement as to how long to treat, particularly with azathioprine once prednisone has been withdrawn. For a few there will be no relapse off all treatment, but over 80% will eventually biochemically relapse [Hegarty et al. 1983]. Some of the factors that predict relapse are a failure to maintain consistently normal transaminases during therapy, time to initial biochemical remission, high IgG at entry and marked portal cell infiltrate [Miyake et al. 2005; Verma et al. 2004]. Most recently Montano-Loza demonstrated that relapse is less likely if the AST, gamma globulins and IgG are normal prior to withdrawal of treatment [Montano-Loza et al. 2007b]. A main factor distinguishing patients who relapse is the duration of therapy preceding withdrawal of immunosuppression: sustained remission was obtained in 67% of patients treated for less than four years prior to cessation of therapy, in contrast to only 10% of those treated for one to two years [Kanzler et al. 2001]. Additionally it is recognized that over 80% of patients on an indefinite azathioprine strategy (2 mg/kg) will remain in remission over a five-year period [Johnson et al. 1995]. Although relapse typically responds quickly and completely to retreatment, such patients are at increased risk of further relapse, and poorer outcomes, if drug withdrawal is reattempted [Montano-Loza et al. 2007a]. If relapse occurs, prednisone will be needed for remission, and azathioprine (or equivalent) is usually given on lifelong basis.
Very few patients fail therapy assuming good compliance. Ethnicity, onset at an early age, hyperbilirubinemia, HLA DRB103 and a presentation MELD score of over 12 [Montano-Loza et al. 2007c; Verma et al. 2007] are important. The choice of intervention in treatment ‘resistant’ patients is arbitrary. As the numbers are small, data are limited, and sometimes extrapolated from studies focusing on new primary steroid-sparing treatments. Some try a higher dose of pre-dnisone but alternative agents, such as calcineurin inhibitors, budesonide, mycophenolate mofetil (MMF) and rapamycin have all been suggested. Cyclosporin has been used successfully as primary treatment in pediatric and adult patients, and can act as a steroid-sparing agent [Cuarterolo et al. 2006]. Similar studies have focused on the related agent, tacrolimus [Larsen et al. 2007; Aqel et al. 2004]. For both agents if chosen the clinician must carefully appraise the risk from renal and neurological toxicity.
The potential for side effects from prednisone has driven interest in alternative steroid agents, particularly budesonide with its 90% first-pass hepatic metabolism (except in cirrhosis or portal hypertension [Danielsson and Prytz, 1994; Hempfling et al. 2003]). A large randomized trial of budeso-nide versus prednisone in new onset noncirrhotic AIH has been presented [Manns et al. European Association for Study of the Liver, 43rd Annual Meeting, 2008]. Patients were randomized for 6 months to either prednisone 40mg/day (40 mg for the first 4 weeks, and then tapered down to 10 mg/day by week 9; tapering was allowed from week 3 if in remission) or budesonide 3mg three times per day (upon biochemical remission budesonide dose was reduced to 3mg b.i.d.). All patients received azathioprine at a dose of 1–2 mg/ kg/day. Biochemical remission was defined as AST and alanine aminotransferase (ALT) within normal range and a complete response as biochemical remission as well as a lack of steroid side effects throughout the treatment period. Budesonide was shown to be superior in achieving biochemical remission over the study period (60% vs 38.8%) with significantly fewer steroid side effects (72% vs 46.6%), and greater efficacy at achieving the primary endpoint of a complete response. However this trial lasted only six months and notably there was no difference in the total number of patients achieving a normal bilirubin or IgG. The dose of prednisone at 40mg is higher than many use and thus whilst this study verifies the principle that budesonide is effective in inducing remission of previously untreated patients (if noncirrhotic), the issues over relevant side effects, long-term therapy, and particularly cost efficacy remain debatable. In contrast, the studies of budesonide for nonresponders have been disappointing, although this population will include cirrhotics [Czaja and Lindor, 2000].
MMF is free of nephrotoxicity with its side effects being similar to azathioprine; in addition some report diarrhea and headaches. For those intolerant of azathioprine there is a role [Chatur et al. 2005], and additionally MMF may be more potent for those with a poor response to therapy [Richardson et al. 2000; Devlin et al. 2004; Czaja and Carpenter, 2005]. Finally analogous to their use in other autoimmune/transplant settings, agents such as the anti-CD20 monoclonal antibody rituximab [Santos et al. 2006] may prove beneficial for selected patients.
Pregnancy is a frequent concern for young women with AIH. Disease flares may occur during and immediately after pregnancy [Werner et al. 2007; Schramm et al. 2006; Buchel et al. 2002; Heneghan et al. 2001] but no factors are predictive, so monitoring is required for all. Management is individualized in terms of the stage of disease and the degree of liver injury. In essence ongoing use of either prednisone and/or azathioprine is considered acceptable. Although azathioprine crosses the placenta there is a lack of teratogenicity in newborns and little evidence to suggest that there should be a dose reduction or drug withdrawal in patients who become pregnant while taking azathioprine. Supplemental or de novo prednisone use may be required but counter intuitively perhaps, most physicians shy away from de novo administration of azathioprine until after delivery.
PBC is a chronic nonsuppurative granulomatous cholangitis of the small interlobular bile ducts, seen most frequently in middle aged women [Gershwin and Mackay, 2008; Kumagi and Heathcote, 2008]. Historic data showed the natural history of histological changes in PBC to be progression to biliary cirrhosis: development of cirrhosis (Ludwig's stage 4 of PBC) over 4 years can be expected in 31%, 50%, and 68% of patients who start respectively from stage 1 (early), 2 or 3 disease [Locke et al. 1996]. Many patients remain asymptomatic whilst others have, or subsequently develop, fatigue and/or pruritus, which can be very distressing. With earlier diagnosis, contemporary series demonstrate most patients have no fibrosis (or symptoms) at presentation. Specific antimito-chondrial antibodies (AMA) are found over 95% of the time, such that the combination of cholestatic liver biochemistry and AMA has a very high predictive power and routine liver biopsy has decreased in popularity [Zein et al. 2003]. An isolated observation of AMA conversely need not equate with PBC, nor be a need for treatment [O'Brien et al. 2008; Leung et al. 2007].
Japanese researchers first isolated UDCA, the 7-ß epimer of chenodeoxycholic acid from bear bile, itself a choleretic in traditional medicine [Makino and Tanaka, 1998] and the prevailing view is that UDCA changes the natural history of PBC [Silveira and Lindor, 2008b; Lindor, 2007]. It is less hydrophobic and more hydrophi-lic than other bile acids, with apparently less toxi-city towards the hepatobiliary epithelium. Its therapeutic mechanism of action includes expansion of the hydrophilic bile acid pool, a direct choleretic effect (bicarbonate-rich choleresis and modulation of biliary transporter proteins), anti-inflammatory effects (interacts with the glucocor-ticoid receptor in vitro) and anti-apoptotic effects on hepatic epithelia. Cessation of therapy with UDCA leads to a rapid return of abnormal liver biochemistry in those with PBC who respond, demonstrating that the initiating and perpetuating factors remain unresolved.
Normally UDCA accounts for about 4% of bile acids but with pharmacotherapy it becomes the predominant bile acid. Over time studies confirmed a correlation between the degree of bile enrichment and improvement in liver biochemistries and PBC Mayo risk score (age, bilirubin and albumin levels, prothrombin time, and presence or absence of peripheral edema), and overall in PBC the data suggest that the effective dosage for UDCA is 13-15 mg/kg per day, commonly, but not necessarily, as a single daily dose [Angulo et al. 1999; Poupon et al. 1997]. Sideeffects are minimal and include weight gain (3 kg in the first 12 months), hair loss, and rarely GI upset [Siegel et al. 2003]. Additionally nothing suggests UDCA is teratogenic but it is usually stopped before and during the first trimester, and restarted subsequently, with there being a good safety profile from its use in intrahepatic choles-tasis of pregnancy.
Many studies have attempted to demonstrate clinical efficacy for UDCA, particularly following work demonstrating a fall in bilirubin values in patients with PBC, with most trials showing beneficial effects of UDCA on biochemical parameters in particular (but not symptoms) [Poupon et al. 1994, 1991, 1987; Heathcote et al. 1994]. With such a slow natural history any individual trial inevitably lacks the power to address endpoints such as death or liver transplantation. The cumulative evidence thus comes from individual and pooled analysis. Of note the Mayo Clinic [Lindor et al. 1996, 1994; Jorgensen et al. 2002] followed up patients with PBC treated with UDCA and have demonstrated a treatment-related prolongation in transplant free survival. French cohorts with long-term follow-up similarly demonstrate survival benefit [Poupon et al. 1999; Corpechot et al. 2005] and this was reiterated by a Spanish cohort including nearly 200 patients treated for a mean of 6.7 years [Pares et al. 2006]. In those with early stage disease, a biochemical response to UDCA (normalization or a 40% reduction from baseline) after 1 year was associated with a similar survival to the matched control population. ter Borg and colleagues in the Netherlands  were able to show a survival benefit without transplant for low-risk patients compared to survival predicted by the Mayo risk model; moreover overall survival was nearly the same as for a control population. A recent publication [Corpechot et al. 2008] looked at defining the biochemical response a year after treatment, to identify UDCA-treated patients at risk of death or transplant. Reiterating the value of a response to UDCA, those with an ALP < 3 ULN and AST <2 ULN and bilirubin ≤17.1μmol/l after 1 year of UDCA had a ten-year transplant-free survival rate of 90% (compared with 51% for those who did not). Histological progression may therefore be halted (but not reversed) by UDCA [Poupon et al. 2003; Corpechot et al. 2000], and studies show that after five years of UDCA treatment the incidence of cirrhosis is only 4%, 12%, and 59% respectively among patients followed-up from Ludwig stages 1 to 3 [Corpechot et al. 2002]. Therapy also appears to reduce the onset and severity of portal hypertension [Lindor et al. 1997].
Overall a combined analysis of three trials (548 patients) [Poupon et al. 1997] showed a one-third reduction in the risk of death or transplant, for patients with moderate-to-severe PBC. However meta-analyses have been inconsistent. Some conclude no beneficial effect on the incidence of death and/or transplant despite improvements in serum bilirubin, jaundice, and ascites [Gong et al. 2008; Goulis et al. 1999]. This contrasts with others showing that long-term treatment with the optimal dose of UDCA could improve survival free of transplant [Shi et al. 2006]. One needs to appreciate the varying, and often inadequate sample size and treatment duration of several studies, as well as inclusion of data from trials using suboptimal doses.
Finally, strong surrogate support for a long-term beneficial effect is available from trends in mortality for PBC [Mendes et al. 2008; Jackson et al. 2007] as well as liver transplantation: for the latter the proportion, as well as absolute number, of patients transplanted for PBC is decreasing [Lee et al. 2007], something not noted for PSC.
Classical immunosuppression has been disappointing. A small three-year randomized control trial of prednisolone showed a significant reduction in the serum bilirubin level; however osteoporosis was an issue [Mitchison et al. 1992]. Benefits are unlikely from short-term therapy and although agents for osteoporosis are available, there has been little enthusiasm for further long-term studies. It was envisaged that budeso-nide could benefit patients with PBC and in the first trial improvement in liver biochemistry, IgM values and liver histology was observed for the few patients investigated (most had early disease) [Leuschner et al. 1999]. A later study (most had advanced disease) found no benefit [Angulo et al. 2000], although subsequent work from Finland has alluded to a histological benefit when combined with UDCA [Rautiainen et al. 2005]. The observation that interface hepatitis on baseline biopsy is predictive of outcome in PBC (as well as a failure to normalize transaminases with UDCA) does suggest a subgroup may benefit from immunosuppression [Corpechot et al. 2004] and interest remains in using budesonide adjunctively in those with more florid hepatitic features.
Other trials of immunosuppressive therapy in PBC have employed azathioprine, cyclosporin and methotrexate in particular. Neither of the two trials of azathioprine showed benefit [Heathcote et al. 1976] [Crowe et al. 1980], and several small trials and one large trial (349 patients) of cyclosporin therapy have been published without good outcome data to support its use [Lombard et al. 1993]. Methotrexate similarly was championed by some studies [Kaplan et al. 2004, 1999] but when studied in randomized controlled studies is ineffective [Combes et al. 2005; Hendrickse et al. 1999].
The group of patients most appropriate for new agents are those who fail to respond to UDCA (around 40%) and a number of agents are worthy of comment. The addition of bezafibrate (which may increase phospholipid translocation into the bile) has been investigated, but the results to date are early and a well-powered study is needed [Iwasaki et al. 2008]. Atorvastatin with many anti-inflammatory properties beyond its lipid lowering effect was recently shown not to be of value for PBC [Stojakovic et al. 2007]. In view of the cloning of a human retrovirus from a cDNA library derived from PBC biliary epithelia cells, antiretrovirals have been investigated, but convincing data is still awaited [Mason et al. 2008, 2004]. More selective immunosuppression (e.g. rituximab) is being studied as well as approaches focusing on fibrosis modulation. Finally other ways of targeting the bile acid pathway are being pursued. Modifications to UDCA (e.g. nor-UDCA) may appeal as well as interfering with nuclear hormone receptors [Zollner et al. 2006]. The Farnesoid X receptor (FXR) is a bile acid-activated nuclear receptor highly expressed in both the liver and gastrointestinal tract. It has a role in regulating bile and cholesterol metabolism and FXR agonists (e.g. novel bile acids or small molecule agonists such as INT-747) may have a role in the future.
Sclerosing cholangitis is a broad description for a group of disorders in which the biliary tree is damaged by chronic inflammation with subsequent fibrosis and obstruction to free biliary drainage [Weismuller et al. 2008; LaRusso et al. 2006]. This scarring leads to beading and dilatation of the bile ducts, cholestasis, secondary biliary cirrhosis and infectious complications. The term is broad as many different injuries result in the same endpoint (Figure 1), hence when using the term PSC, it is assumed that secondary causes are absent [Abdalian and Heathcote, 2006].
The original descriptions of PSC required clear exclusion of obstruction by stones or malignancy [Chapman et al. 1980], and prior to readily available cholangiography, surgery was often involved in evaluation. Although percutaneous biopsy may show the diagnostic periductal sclerosis, the patchy nature of disease means more often nonspecific biliary changes are described. Radiological assessment of the biliary tree therefore remains clinically the most common diagnostic modality, and this has until recently been by endoscopic or percutaneous cholangiography. With increasingly sophisticated MRI there is a noninvasive and safe imaging modality that is now the initial investigation of choice [Meagher et al. 2007]. This is benefiting those with inflammatory bowel disease (IBD) who still represent a significant cohort at risk for PSC, even when they have normal liver biochemistry: a recent study found MRCP evidence of disease in 20% [Bungay et al. 2008]. Small duct PSC (a variant that progresses over time to large duct PSC in 20% of individuals [Bjornsson et al. 2008]) is the term used for patients with classic histological findings but a normal cholangiogram. Its presence supports broadly the concept of an evolving disease: (a) small duct cholangitis, (b) progressive cholestasis, (c) cirrhosis and (d) decompensation.
In the classic descriptions, the majority of patients had ulcerative colitis and were more often men presenting in their 40s [Bambha et al. 2003]. Symptomatically fatigue, right upper quadrant pain, pruritus and jaundice are common and since the clinical course of PSC is typically one of insidious worsening of cholestasis, end-stage liver disease inevitably develops; prior natural history studies show average survival at 12 to 17 years from symptom onset [Kim et al. 2000]. Bacterial cholangitis, especially after intervention, is seen in about one in three patients, and cholangiocarcinoma develops in as many as 10–20% of patients (commonly presenting within the first year of diagnosis of PSC [Bergquist et al. 2002]).
In interpreting studies, one must first recall that the inclusion criteria have traditionally been stringent with classically all three of abnormal biochemistry, cholangiography and histology being required. This skews enrollment. Although alternative etiologies have been sought, this refers to the traditional etiologies: chronic obstruction by stones, vascular disease (portal vein thrombosis, hepatic artery loss), recurrent cholangitis and cholangiocarcinoma. It has not included IgG4 associated sclerosing cholangitis a steroid-responsive disease. Endpoints for treatment intervention are troublesome with the dilemma of including patients with such advanced disease that efficacy of any treatment is unlikely. As a surrogate outcome measure, the Mayo PSC risk score has frequently been applied, as it predicts the probability of survival (based on age, bilirubin, albumin, AST and the presence/absence of variceal bleeding). Lastly assessing the stage, and progression, of disease by use of serial liver biopsy, as has been common, is limited by sampling errors.
The use of UDCA was naturally studied after the successful studies in PBC, and remains a popular agent [Silveira and Lindor, 2008a]. Indeed its use demonstrates how prescribing habits are a balance of evidence and perceived benefit, and that modern medical practice can be as ‘unscientific’ as traditional medicine.
Biliary enrichment of UDCA increases with increasing dose and reaches a plateau at a daily dose of 22–25 mg/kg in the bile of patients with PSC [Rost et al. 2004]. Various trials investigated UDCA treatment in PSC after Hayashi's and Chazouilleres’ first open label studies in, 1990 [Chazouilleres et al. 1990; Hayashi et al. 1990]. The initial double-blind placebo controlled trial of standard dose UDCA by Beuers and colleagues  demonstrated an improvement in serum liver tests and a histological score. Similar results were described by Stiehl and colleagues . A larger controlled trial from the Mayo with longer follow up confirmed the effects of UDCA on serum liver biochemistry, but without a survival benefit. This important effort was randomized, double-blinded (UDCA at 13–15 mg/kg/day vs placebo) and involved 105 patients (half with advanced PSC), with a median follow up of 2.2 years [Lindor, 1997]. Most subsequent studies have similarly confirmed an improvement in serum liver biochemistry but not more tangible endpoints.
Harnois assessed the tolerability and effectiveness of UDCA at 25–30 mg/kg/day in a pilot study of patients with PSC [Harnois et al. 2001]. Changes in the Mayo risk score after one year of treatment and projected survival at four years of treatment were compared with prior data. High-dose UDCA was well-tolerated, associated with improvement in liver biochemistries, and led to a significant improvement in the Mayo risk score. Other encouraging results with high-dose UDCA were seen in a two-year double-blind, placebo-controlled study conducted by Mitchell and colleagues in which 26 patients with PSC were randomized to 20mg/kg/day of UDCA or placebo [Mitchell et al. 2001]. Further work recently published [Cullen et al. 2008] also points towards benefit from high-dose therapy and this is similar to a, 2005 randomized, placebo-controlled study by Olsson and colleagues (UDCA at 17–23 mg/kg/day) [Olsson et al. 2005]. They found significantly improved liver biochemistry but no significant difference for the combined endpoint of death or liver transplantation.
Thus, overall the studies to date support an effect of UDCA on liver tests, demonstrate no harm, but lack robust data for the hard endpoints patients want to see. Other considerations are relevant. PSC is a premalignant condition and some studies suggest UDCA may counteract this. PSC is associated with a 10–20% lifetime risk for the development of cholangiocarcinoma, and patients with ulcerative colitis and PSC are at a higher risk for colonic dysplasia and colon cancer (0.5–1% per year of disease, explaining the recommendation for regular colonoscopy in those with IBD and PSC, including post-transplant). Gallbladder malignancy is also a concern [Buckles et al. 2002]. There are a few publications supporting the notion that UDCA use is associated with a lower prevalence of colonic neoplasia in patients with PSC and concomitant ulcerative colitis, and this parallels in vivo experimental data [Rudolph et al. 2007; Pardi et al. 2003].
Numerous other agents (e.g. prednisone, bude-sonide, azathioprine) have been used with the hope of improving medical care in PSC, but without positive outcomes. The combination of an antibiotic with UDCA is very attractive and a study with 80 patients with PSC randomized to either UDCA alone or the combination of metro-nidazole and UDCA did find that combination therapy improved serum alkaline phosphatase and Mayo Risk Scores, but without significant effect on disease progression compared to UDCA alone [Farkkila et al. 2004]. In view of the possibility that biliary infection may contribute to the cholangitis, trials of antibiotics continue to be of interest – for example, vancomycin [Davies et al. 2008] and azithromycin [Boner et al. 2007]. Other possible treatments include antifibrotics such as rapamycin or angiotensin converting enzyme inhibitors. The role of cystic fibrosis transmembrane conductance regulator (CFTR) as an important ion channel in biliary biology has led to an ongoing trial of docosahex-aenoic acid which improves CFTR function [Blanco et al. 2004]. Disappointedly despite its role in IBD, tumor necrosis factor alpha (TNF alpha) inhibition does not seem to directly help PSC [Hommes et al. 2008]. Perhaps the most exciting agent at present is an inhibitor of toxic bile formation, 24-norursodeoxycholic acid (a homologue of UDCA with one fewer methylene group in its side chain). Multidrug resistance gene 2 knockout mice represent a well-characterized model for sclerosing cholangi-tis and the novel bile acid 24-norursodeoxycholic acid appears to ameliorate the cholangiopathy [Fickert et al. 2006].
A few important concepts merit discussion. Bacteriobilia is found in the majority of patients with PSC, but clinically apparent cholangitis does not commonly manifest unless intervention has occurred, or the disease has advanced significantly. There is therefore a move away from ERCP unless it is with clear therapeutic intent (i.e. dominant stricture, in which case repeated balloon dilatation may be better than stenting [Kaya et al. 2001]) or there is a raised suspicion for malignancy. Biliary surgery in patients with PSC, other than simple cholecystectomy (surprisingly frequent [Said et al. 2008]), should be minimized and reserved for the selected rare noncirrhotic patients who have marked cholesta-sis or recurrent cholangitis caused by a dominant extrahepatic or hilar stricture not amenable to endoscopic therapy. PSC is an important indication for liver transplantation, reflecting the present inadequacies of medical treatment [Schreuder et al. 2008]. The use of MELD scores is appropriate but some specific circumstances including recurrent bacterial cholangitis despite therapy, severe extrahepatic biliary obstruction that precludes operative repair, and uncontrolled peristomal variceal bleeding may not be adequately addressed by MELD. Post-transplant outcome is good for PSC but recurrent disease is worrying at up to 20% by five years, and may be a greater concern in living related donation. Of the many proposed risk factors for recurrent PSC, the presence of an intact colon in those with colitis is one that has potential for modification, by appropriately timed colectomy [Vera et al. 2002].
Atypical presentations of syndromes including sclerosing cholangitis have been recognized for some time [Toosi and Heathcote, 2004] and the Japanese-defined steroid responsive autoimmune pancreatitis (AIP) that, in its spectrum, includes a sclerosing cholangiopathy (AIP/SC) [Kamisawa and Okamoto, 2008]. Elevated serum IgG4 levels were found to be associated with sclerosing pancreatitis and treatment with corticosteroids decreased levels [Hamano et al. 2001]. Thus, it is now recognized that IgG4-associated cholangi-tis is the biliary manifestation of a multisystem fibro-inflammatory disorder in which affected organs have a characteristic lymphoplasmacytic infiltrate rich in IgG4-positive cells [Bjornsson et al. 2007; Gardner and Chari, 2008]. Specific diagnostic criteria for AIP have emerged over time and one recent example is based on histolo-gic, imaging, serologic characteristics and extra-pancreatic involvement as well as a response to corticosteroids [Chari, 2007]. With greater awareness of disease, it is also clear that IgG4 levels can be normal in a proportion of patients. Other organs are often involved in AIP and their involvement may be diagnosed before, simultaneous with, or after the diagnosis of AIP – for example, in 64 patients who had AIP [Hamano et al. 2006] the most frequent extrapancreatic lesion was hilar lymphadenopathy (80.4%), followed by extrapancreatic bile duct lesions (73.9%), lacrimal and salivary gland lesions (39.1%), hypothyroidism (22.2%), and retroper-itoneal fibrosis (12.5%). Both intrinsic (tubuloin-terstitial fibrosis) and extrinsic (hydronephrosis secondary to retroperitoneal fibrosis) renal disease have been associated with AIP, as has inflammatory pneumonitis and inflammatory pseudotumor of the liver. Biliary tract involvement of both intra- and extrahepatic bile ducts is seen, with the distal common bile duct being the most common site. Patients may also present following surgical exploration for presumed malignancy that turns out to be benign.
It is likely that a proportion of patients previously labeled as PSC have a variant of this disease, suggesting a subgroup that might require different therapy. Furthermore, is the quandary as to the observation that 9% of patients enrolled into a clinical trial of PSC were retrospectively found to have elevated IgG4 levels and these patients appeared to fare worse [Mendes et al. 2006]. AIP/SC differs from PSC in that there is generally less intrahepatic involvement and segmental strictures can be transient; IBD, which is present in two-thirds of those with PSC, is less common (6%) in AIP/SC.
This disease can be spontaneously relapsing and remitting, but the use of steroids (Table 5) predictably induces remission including resolution of pancreatic ductal abnormalities or masses and of extra-pancreatic manifestations (biliary strictures, retroperitoneal fibrosis, salivary gland enlargement). Adjunctive treatment by biliary stenting may still be needed. Whether to treat PSC patients found to have a raised IgG4 level (without other features of AIP/SC) is not resolved nor is there is consensus on steroid regimens or the duration of treatment [Church and Webster, 2008; Ghazale and Chari, 2007]. Starting prednisone doses range from 30 to 40mg in most studies, although lower doses may be effective. An initial high-dose therapy is usually given for three to four weeks followed by a reducing course – for example, prednisone 40mg daily for four weeks followed by tapering of 5mg per week. Such a course allows for assessment of response to therapy, which is typically fast with significant radiological improvement at around two weeks; a fall in serum IgG4 levels and liver enzymes also guides response. In patients who experience a complete radiological response, normalization of imaging findings typically occurs within just over one month. Where needed biliary stent removal can usually occur within 2 months of treatment, although if there are fibrotic changes in the biliary tree or retro-peritoneally radiological resolution may be limited. Recent data suggest that relapse is likely in 70% of patients who have proximal extrahepatic or intrahepatic biliary strictures. Chronic immu-nosuppression achieved by the use of long-term low-dose steroids or by addition of immunomodulatory medications such as azathioprine or mycophenolate mofetil may therefore be appropriate and rituximab may be an option for a few [Ghazale et al. 2008; Topazian et al. 2008; Church et al. 2007; Hirano et al. 2007].
In predicting the emerging treatments for this field, it is hoped that the combined efforts of genetic analysis (association studies, gene expression, copy number variation, methylation), proteomics, imaging and animal models will lead to new drug targets. The slow natural history exhibited by these diseases and the fact that patients are being diagnosed at much earlier stages of disease will remain challenges for those designing studies. Regardless, the prospects for avoiding liver failure is excellent for those diagnosed today with autoimmune liver disease.
Both GM Hirschfield and EJ Heathcote have received research support from Axcan Pharma, and are recruiting patients to an ongoing trial of INT-747 supported by Intercept Pharmaceuticals.
Gideon M. Hirschfield, Toronto Western Hospital, Toronto, Canada ; Email: firstname.lastname@example.org.
Nadya Al-Harthi, Liver Centre, Toronto Western Hospital, University Health Network, University of Toronto, Toronto, Canada.
E. Jenny Heathcote, Liver Centre, Toronto Western Hospital, University Health Network, University of Toronto, Toronto, Canada.