Review article: overlap syndromes and autoimmune liver disease

Authors


  • This commissioned review article was subject to full peer-review.

Correspondence to:

Dr G. M. Hirschfield, Centre for Liver Research and NIHR Biomedical Research Unit, 5th Floor, Institute of Biomedical Research, The Medical School, University of Birmingham, Birmingham B15 2TT, UK.

E-mail: g.hirschfield@bham.ac.uk

Summary

Background

Autoimmune hepatitis (AIH), primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) all nestle within the family of autoimmune liver diseases, whereby the result of immune-mediated liver injury gives rise to varied clinical presentations. Some patients demonstrate a phenotype whereby there is evidence of either PBC or PSC together with overlapping features of AIH. Due to an absence of well-validated diagnostic criteria and a lack of large therapeutic trials, treatment of overlap conditions is empiric and extrapolated from data derived from the primary autoimmune liver diseases.

Aims

To review overlaps in the context of autoimmune liver diseases.

Methods

General and specific review of published articles using PubMed, Medline and Ovid search engines, alongside pre-existing clinical management protocols, guidelines, and the authors' own knowledge of the published literature.

Results

The challenges in diagnosis, clinical presentation, determining natural history and outcome of overlaps are presented, as well as present-day management suggestions, some based on evidence, others on consensus and opinion.

Conclusions

Overlapping autoimmune features, be they clinical, serological, histological or radiological are not infrequent, but appropriate diagnosis remains hindered by a lack of standardised diagnostic criteria. Optimum care for those with suspected overlap should thus focus on attention to detail over the fundamental aspects of timely secure diagnosis of the dominant disease entity. Clinicians should counsel patients carefully with regard to the risks and benefits of treatment, bearing in mind the paucity of randomised and controlled outcome data for medical interventions.

Introduction

Autoimmune hepatitis (AIH), primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) are characterised by a varying degree of immune-mediated liver injury, and although criteria exist which facilitate their timely diagnosis, shared serological, immunological and histological patterns exist across the spectrum of these three autoimmune liver diseases (Table 1). Conditions exhibiting features of two different autoimmune liver diseases are commonly designated ‘overlap syndromes’; however, there is no current agreement on what constitutes an overlap, neither are there specific diagnostic criteria, suggesting that overlaps are not distinct entities as such. Thus, rather than representing a distinct disease process, these patterns of immune-mediated liver injury are likely a manifestation of the inherent distribution of clinical phenotypes across patient populations with immune-mediated hepatitis (interface activity) forming a fundamental component of most clinical overlap presentations. In this article, we provide an update on overlap syndromes, discussing the diagnostic and therapeutic challenges, which exist across the varied clinical presentations encountered in routine practice.

Table 1. The phenotype of autoimmune liver diseases
 AIHPBCPSC
  1. AIH, autoimmune hepatitis; AiSC, autoimmune sclerosing cholangitis; AMA, antimitochondrial antibody; ANA, antinuclear antibody; ASMA, anti-smooth-muscle antibody; IBD, inflammatory bowel disease; LKM, liver–kidney microsomal; MRCP, magnetic resonance cholangiography; OADC, oxoacid–dehyrogenase complex; pANCA, perinuclear anti-neutrophil cytoplasmic antibody; PBC, primary biliary cirrhosis; PDC-E2, E2 component of pyruvate dehydrogenase; PSC, primary sclerosing cholangitis; SLA/LP, soluble liver/pancreas antigen; ULN, upper limit of normal.

  2. a

     Disease onset tends to be bimodal with a first peak incidence ~age 10–20 and a second peak ~age 45. However, up to 20% of all AIH patients are above the age of 60.[107-110]

  3. b

     Cholangiography is normal in small-duct disease.

  4. c

     In PBC, a degree of parenchymal inflammation may be observed secondary to bystander, collateral hepatocellular injury. In such cases with ‘florid’ interface hepatitis, there is anecdotal data demonstrating the efficacy of immunosuppression, particularly budesonide.[67, 111] However, these studies are not yet supported by well-controlled, double-blind, randomised trials.

Age at diagnosisAll agesa Middle age (>45 years)All ages (onset usually ~40 years)
Gender predominanceFemale > Male (Type I – 4:1; Type II – 9:1)Female > Male (9:1)Male > Female (7:3)
Serology
ANA>1:40 titre in 70–80% (often with ASMA)Disease specific ANA exist in ~30–50% Nonspecific ANA exist in 70–80%
ASMA>1:40 titre in 70–80% (often with ANA)May be presentUp to 83%
Anti-LKM-13–4% (categorise AIH-II)  
Anti-SLA/LP10–30%May be presentMay be present
pANCAUp to 92% 26–94%
AMALow titre in ~5–10%~95% (anti-PDC-E2 highly specific)Coincidental if present
Specific autoantigen

CYP IID6 specific autoantigen for LKM-1

F-actin specific for ASMA

PDC-E2 (and E2 components of other OADC proteins) specific for AMANot identified
Immunoglobulins

Elevated IgG (>1.2xULN)

IgA deficiency detected in AIH-II

IgM elevated in most

IgG elevated in 61%

IgM elevated in 45%

MRCPUsually normal although mild intrahepatic ductal changes can be seen in advanced fibrosis (10%) Cholangiographic changes can be seen in up to 50% of children (so called AiSC)NormalDiagnostic: multifocal stricturing throughout the hepatobiliary tree b
Liver histology
Interface hepatitisCharacteristic findingVariably presentVariably present
Portal inflammationLymphaplasmacytic infiltrateLymphocytic infiltrateLymphocytic infiltrate
Biliary changes~10% of casesInflammatory duct lesionClassically: onion-skin' periductal fibrosis
GranulomasNoCharacteristic (but only present in a few cases)Atypical (<10%)
Coexisting IBD

~3–10%

PSC should be excluded

Not characteristic~80% (geographically variable)
Response to immunosuppressionYesNoc No

Shared Aetiopathogenic Mechanisms

Broadly similar mechanistic themes of injury are described for AIH, PBC and PSC, whereby liver disease likely represents the result of a cell[1-6] and antibody-mediated[7-10] immunological attack against liver-specific targets.[11-13] The precise factors leading to disease initiation and perpetuation are unknown, but likely reflect a combination of genetic predisposition[14-17] relating to defects in immunological control of autoreactivity, as well as environmental triggers,[18-24] which precipitate a persistent breakdown in self-tolerance. Despite genetic susceptibility loci being strongly linked to the HLA region, these are generally distinct across all three disease entities.[25-27] Furthermore, there is barely any significant genetic overlap between PBC and PSC. In AIH, immune-mediated liver injury is most pronounced in the portal/periportal areas, and although a degree of parenchymal damage can be observed in PBC and PSC, the pattern of disease in these two conditions is predominantly directed toward biliary epithelial cells. Hepatitic inflammation in the latter is likely mediated via bystander mechanisms, which cause collateral damage to hepatocytes, subsequent release of intracellular antigens, and the development of autoantibodies, but this itself is speculative (Figure 1).

Figure 1.

Aetiopathogenesis of autoimmune liver disease. The factors leading to biliary-predominant or hepatic-predominant tissue damage are not certain, although there likely exist common aetiopathogenic factors across all three conditions. In a genetically predisposed individual, an environmental trigger may lead to development of neo-autoantigens, which are recognised by nonpolarised T cells (via an antigen-presenting cell) resulting in cellular activation. Up to 80% of PSC patients (and ~4.5% of patients with AIH) have coexisting IBD, and aberrant homing of memory lymphocytes from bowel to the liver; or bacterial translocation from an inflamed ‘leaky’ gut may explain this association. Once activated, and depending upon the cytokine milieu, activated intrahepatic T-helper cells promote the differentiation of cytotoxic T cells, Th17 cells, and/or the differentiation of B cells into plasma cells which produce immunoglobulins. Deficiencies in the number and function of peripheral blood regulatory T cells (T regs) have also been demonstrated in AIH and PBC (the role of T regs in PSC being less clear), and together with the activation of monocytes contributes to the loss of self-tolerance and perpetuation of chronic inflammatory tissue damage.

Serological Patterns of Reactivity

Serum autoantibodies are frequently identified in autoimmune liver diseases, and although their presence is neither necessary for diagnosis, nor generally prognostic, the pattern of serology is commonly used to subclassify disease clinically. Type-I AIH is characterised by anti-nuclear antibodies (ANA) and/or anti-smooth muscle antibodies (ASMA), whilst type-II AIH is characterised by anti-liver kidney microsomal type-1 (anti-LKM-1) antibodies. The predominant antigen in type-II AIH has been identified as human cytochrome P450IID6.[28, 29] Anti-soluble liver–pancreas antigen (SLA/LP) antibodies (directed against a selenocysteinyl-transfer RNA synthase)[30] were historically thought to represent a third subgroup of AIH patients; however, over 75% of anti-SLA/LP-positive individuals are also ANA and/or SMA positive, being clinically indistinguishable from those with type-I disease.[31]

Anti-mitochondrial antibodies (AMA) are the serological hallmark of PBC and detected in ~95% of such patients. Positive ANA titres are also found in 30–50% of individuals with PBC (more commonly in the few who are AMA negative), but in this setting ANA reactivity is, in contrast to AIH, often antigen specific (anti-gp210 and anti-sp100).[32] Patients with AIH histologically may also be AMA positive (~10%), generally behaving like typical AIH, albeit with rare instances of transition to PBC.[33] Conversely, up to 19% of patients with AIH, and ~5% of patients with PBC can present with seronegative disease at the time of diagnosis.[34] Seronegative AIH follows a similar course to autoantibody-positive disease, and falsely classifying patients as having cryptogenic chronic hepatitis may delay the institution of appropriate treatment.[35] The same holds true for AMA-negative PBC. Atypical, nonspecific antibodies to neutrophil cytoplasmic antigens (ANCA), distinct from those seen in microscopic polyangitis or Wegener's granulomatosis, are detectable in up to 88% of patients with PSC,[36] ulcerative colitis (~87%) and AIH (50-96%). Although ANCA titres correlate with disease activity in the systemic vasculitides, this is not observed in autoimmune liver disease or IBD.[37-39] ANA (8–77%) and ASMA (up to 83%) reactivity is also variably seen in PSC.[40]

Newer, potentially more specific autoantibodies (Table 2) are increasingly being identified although testing for many of these is not widely available, and their diagnostic and prognostic value has yet to be firmly established. The mere presence of autoantibodies is not synonymous with autoimmune liver disease, as ANA can also be detected in non-alcoholic fatty liver disease (21%), asymptomatic blood donors (26%), pregnancy (10%), malignancy (30%) and up to 50% of individuals with infectious disease.[41, 42] ASMA may also be detectable in ~43% of normal healthy individuals. Thus, although clearly useful in the workup of autoimmune liver diseases, autoantibodies are not to be used in isolation, rather interpreted in conjunction with the spectrum of radiological and histological investigations performed.

Table 2. Autoantibodies in autoimmune hepatitis and overlap syndromes[42, 112]
AutoantibodyAntigenic targetComments
  1. AMA, anti-mitochondrial antibody; ANA, anti-nuclear antibody; ASMA, anti-smooth muscle antibody; dsDNA, double-stranded DNA; LKM, liver–kidney microsomal; pANCA, perinuclear anti-neutrophil cytoplasmic antibody; PDC-E2, E2 component of pyruvate dehydrogenase; SLA/LP, soluble liver/pancreas antigen.

  2. a

     Low-titre ANA or ASMA (1:10–20) are always considered abnormal in children.

ANAVarious nuclear antigens: nuclear membranes, DNA, centromeres, ribonucleoproteins and cyclin A

Characterise AIH-I where they are often present in high titres (>1:160), although levels do not correlate with disease severity.a

Present in isolation (13%) or with ASMA (54%)

ASMAVarious: actin, vimentin, tubulin, desmin

Characterise AIH-I where they are often present in high titres (>1:160), although levels do not correlate with disease severity.a

Present in isolation (33%) or with ANA (54%)

50% also positive for anti-F-actin

Anti-LKM-1CYPIID6

Characterise AIH-II

Almost always associated with disease relapse on corticosteroid withdrawal

Anti-LKM-1 positive individuals present more frequently with fulminant hepatic failure (25%)

Anti-SLA/LPSep (o-phosphoserine) tRNA:Sec (selenocysteine) tRNA synthase

Detectable in both patients with AIH-I and II

May be present without other conventional antibodies in 10-30% of cases

Frequently associated with more severe histological damage, the need for long treatment durations, relapse after drug withdrawal, treatment dependence, liver related death or transplantation

Associated with HLA DRB1*0301

Anti-actinF-actin and α-actinin

Anti-F-actin more specific (80%) and sensitive (90%) than ASMA in AIH

Associated with treatment dependence in children and poor liver-related outcomes

The presence of anti-α-actinin antibodies at baseline may help predict treatment response

Cross-reactivity of anti–α-actinin antibodies against actin or ssDNA associated with severe clinical and histologic disease

Anti-liver cytosol-1Formiminotransferase cyclodeaminase

Associated with type II AIH

Can be present when anti-LKM-1 absent

pANCAVarious

Nonspecific

Present in higher titres in AIH than seen in PSC

AMAPDC-E2

Highly specific and sensitive for PBC

Consider potential misread of immunofluorescence for LKM-1 antibody if found in AIH

Anti-dsDNADNA

Positive in:

Up to 60% of AIH/PBC-overlap cases

4% of PBC patients

26% of individuals with AIH

Anti-asialoglycoprotein receptorAsialoglycoprotein receptor

Linked with increased hepatocytic apoptosis and persistent interface hepatitis in AIH

May be associated with relapse after treatment withdrawal

Clinical Presentation of Autoimmune Liver Disease

Although broadly similar immunological pathways lead to the development of AIH, PBC and PSC, there exist distinct differences in the age at which disease first presents, gender predisposition and clinical phenotype (Table 1).[43-45] Overlap presentations seem more common in younger individuals with AIH (e.g. see later for discussion on autoimmune sclerosing cholangitis), although no systematic studies exist to comment on the average age of overlap presentations because of their relative rarity and absence of definitive/agreed ‘diagnostic’ tests. In addition, there is inadequate knowledge of the impact of ancestry on rates of overlap presentation and a need for more definitive surveys in this regard. The clinical presentation in AIH can vary from acute, fulminant hepatitis and liver failure,[46-51] to the patient who presents with established cirrhosis (up to 30%)[52] and decompensation. Seemingly acute presentations are not really a feature of PBC and PSC, being reported only in small anecdotal reports, and clinically often being hard to distinguish from superimposed drug injury or late presentation of patients.[53]

Symptoms are highly nonspecific when present in autoimmune liver disease, and include fatigue, arthralgia and pruritus; the latter being more common in PBC and PSC. Increasingly, patients are being identified with autoimmune liver disease when incidental abnormal liver biochemistry tests are followed up, and who are otherwise asymptomatic. All three conditions are frequently associated with extrahepatic autoimmune diseases (up to 20%) either at presentation or during follow-up. Moreover, the majority of patients of European descent with PSC have coexisting inflammatory bowel disease (IBD), usually ulcerative colitis (up to 85%),[54] although IBD can occur in AIH, albeit with a much lower frequency (4.5–18%[55, 56]).

Diagnosis

There is no single test for AIH, and diagnosis requires interpretation of autoantibody profiles, serum biochemistry, liver histology and perhaps most importantly, the exclusion of other disease aetiologies.[57] This feeds into the absence of diagnostic tests for overlap syndromes, in which AIH is always a central theme, be the principal disease process PBC or PSC. The predominant biochemical feature in AIH is of a transaminitis, which can vary from minor elevations of AST and ALT to >10xULN in 50% of patients.[58] A significantly elevated ALP or γGT should prompt investigation for coexisting biliary disease. Another characteristic hallmark of AIH is a marked elevation of IgG titre, particularly when compared to either IgA or IgM. Coexisting elevation of both IgG and IgM can be observed in AIH/PBC-overlaps, but elevation of all three immunoglobulin subclasses is often a nonspecific sign of advanced cirrhosis. Histology is vital in evaluating those with AIH and overlap presentations, although experts disagree as to its routine utility because there are no specific features. The most frequent histological finding in AIH is a lymphoplasmacytic infiltrate (Figure 2a), predominantly in the periportal area invading the limiting plate and primarily characterised by interface hepatitis (Figure 2b). However, interface hepatitis can be also seen in approximately 25% of patients with PBC and in PSC.[59, 60] Moreover, some bystander biliary injury is observed 10–20% of patients with AIH (Figure 2c,d),[61] and not synonymous with a diagnosis of a distinct PBC or PSC-overlap syndrome. Biliary changes in AIH need to be interpreted on a case-by-case basis in combination with biochemical indices, serological markers, clinical presentation and response to immunosuppression. Whilst sampling variability must always be borne in mind, the true value of advanced liver biopsy interpretation is in determining the predominant pattern of immune mediated liver injury and, particularly where serial biopsies are available, the pattern of fibrosis progression in the context of biochemical activity and treatment.

Figure 2.

Histological features of autoimmune hepatitis. (a) Plasma cell-rich hepatitis in acute AIH (H&E ×100). (b) Classical interface hepatitis in AIH (H&E ×200). (c) Cytokeratin 7 staining demonstrating biliary injury in acute AIH. (d) Masson trichome staining demonstrating marked interface hepatitis and bystander bile duct injury. Adapted with permission: Sherlock's Diseases of the Liver and Biliary System 2012.[138]

The combination of positive AMA and an elevated ALP, with or without elevated IgM levels and symptoms of cholestasis, is usually sufficient to make a diagnosis of PBC. Histological features are of a chronic, nonsuppurative, destructive cholangitis of the small interlobular bile ducts, although liver biopsy is not generally advocated except for prognostication purposes, or in rarer cases where PBC is suspected on clinical grounds and the AMA is undetectable (~5%). In this regard, the reduction in frequency of liver biopsy use in PBC raises the concern that patients with interface hepatitis are being overlooked. The natural history of AMA-negative PBC is similar to that of AMA-positive disease; however, the presence of ANA is nearly universal in the former, with specific patterns on immunofluorescence being characteristic in PBC.[32] In PSC, laboratory investigations often reveal a cholestatic biochemical profile, but liver enzymes fluctuate during the course of disease, and normal levels should not preclude further diagnostic steps if PSC is suspected on clinical grounds. Accurate diagnosis is obtained through cholangiography (usually MRCP) demonstrating multifocal strictures throughout the biliary tree. In some cases, liver biopsy is invaluable, particularly for determining small duct disease in the presence of normal large duct cholangiography as well as some estimate of disease stage, severity of fibrosis and evaluating for overlapping immune-mediated hepatitis.

The Clinical Challenge of Overlap Syndromes

The imprecision of immune-mediated liver injury means that ‘overlap features’ be they biochemical, serological, histological or radiological, are frequently observed across the classic autoimmune liver diseases (Figure 3). The term ‘overlap syndrome’ is applied to describe poorly defined instances where either concurrently or consecutively there exists a coexistence of AIH, as well as clear features of either PBC or PSC. The challenge remains that no autoimmune liver disease has an absolute diagnostic test (the possible exception being PBC), all being diagnosed based on the presence and relative absence of various markers of clinical, biochemical, serological, radiological and histological disease, with some clearly being less categorical and objective than others. This appraisal must be performed longitudinally rather than at a single point in time. Overlap syndromes therefore likely represent rather than a distinct process, the inherent distribution of clinical features across patient populations; the more extreme the distribution, the more distinct the apparent overlap. The prevalence of overlap features is hard to ascertain because of publication bias, challenges in definitions (serological overlap is arguably not of the same significance as histological or radiological overlap), and limitations in test interpretation (e.g. there are insufficiently reproducible ways to grade interface hepatitis, and interface hepatitis itself is likely a common mechanism of liver injury across diseases). Overlap designations therefore tend to be arbitrary and imprecise, and the clinical phenotypes of patients with the same overlap designation exhibit considerable heterogeneity.[62, 63] Presentations that raise the spectre of overlap therefore span:

  1. An immunoserological overlap: e.g. positive ANA/ASMA titres and elevated IgG in conjunction with AMA-positive PBC; or AMA positivity in AIH;
  2. A biochemical overlap: AST/ALT >5xULN in patients with PBC or PSC; or ALP >3xULN in patients with AIH (or γGT >5xULN in children);
  3. A radiological overlap: clinical features of AIH with cholangiographic abnormalities indicative of an inflammatory cholangiopathy;
  4. A histological overlap: lymphoplasmacytic infiltrate and interface hepatitis on liver biopsy with bile-duct lesions indicative of either PBC or PSC;
  5. Varying combinations of the above including temporally i.e. consecutive vs. sequential presentations.
Figure 3.

The spectrum of autoimmune liver disease. Overlap features are frequently observed between the classic autoimmune liver diseases, a likely reflection of the inherent distribution of clinical features across the three core autoimmune liver diseases. In this pictographic representation, interface hepatitis has been used as an example. In AIH, the majority of patients will have a significant degree of interface hepatitis; however, a degree of interface hepatitis can be also observed in a proportion of patients with PBC and PSC. Approximately 10% of patients with AIH may also have histological features of bile duct injury. A similar graph can be adopted for other parameters such as autoantibody titre, serum immunoglobulins and liver biochemistry e.g. 5–20% of patients with AIH have anti-mitochondrial antibodies (AMAs), 15% have increased serum levels of IgM and 19% have a disproportionate elevation of serum ALP.

With no codified diagnostic approach, reported prevalence figures are variable, with some clinicians identifying an overlap as infrequently as 5%, whilst others see patients with overlap syndromes as often as 20% of the time in patients with established AIH. It is the authors' contention that overlap syndromes should be diagnosed conservatively and robustly, and that clinical investigations have different significance in the diagnostic pathway, with a good-quality liver biopsy and/or cholangiogram presenting the strongest means to diagnose overlap. Clinically overlap should be considered in the differential when a patient deviates from the normal clinical course and expected response to therapy, but it is not necessary to over diagnose. For example, when patients in a clinical trial of PBC were identified as having potential AIH overlap retrospectively, it was notable that the clinical course on ursodeoxycholic acid (UDCA) treatment was not distinct as compared with classical PBC patients, nor was their outcome.[64] Clinicians must also be mindful of the varied presentation of drug-induced liver injury and the potential for clinical changes to be related to toxicity from prescribed and nonprescribed agents.

AIH/PBC overlap

There is rarely any confusion about making a definitive diagnosis of PBC; however, approximately 10% of patients with all the features of AIH may also be persistently AMA positive. This in itself is not synonymous with a distinct AIH/PBC-overlap syndrome and in a cohort study of AMA-positive patients with AIH (n = 15), none had any histological features suggestive of PBC.[65] Moreover, those patients who were treated conventionally with steroids did not show clinical or histological evidence of PBC despite the continued detection of AMA over a 27-year follow-up.

The concept of AMA-positive PBC with true overlapping features of AIH has arisen as a degree of parenchymal inflammation akin to that observed in AIH is increasingly recognised in PBC, usually in the form of interface or lobular hepatitis. In rare circumstances, this may be more pronounced than the cholestatic component with ALT or AST values as high as 5xULN. In this event, disease progression is related in part to the severity of interface activity.[66-68] The most readily identified presentation in such a situation is the simultaneous presence of both diseases although less commonly, the onset of AIH and PBC is temporally dissociated, usually with PBC presenting first,[69] having a variable interval of 6 months to 13 years before the onset of AIH.[70] A formal change in diagnosis from one condition to the other has also been reported, although this may take many years to develop.[33] There do not appear to be any significant differences in the acuity of symptom onset among patients with AIH/PBC overlap vs. AIH alone.

Extensively validated, stringent criteria are lacking, although the classification proposed by Chazouillères et al. remains the most commonly used tool for diagnosing AIH/PBC overlap.[71] Some clinicians include more ‘flexible’ criteria for diagnosing AIH/PBC overlap, such as an improvement in liver biochemistry whilst on immunosuppression in patients already diagnosed with PBC.[72] The reported prevalence in the literature of AIH occurring in the context of PBC (depending on the criteria used) varies between 2.8% and 19%. Up to 25% of these patients are ASMA positive, whereas true positivity for anti-LKM1 antibodies is seldom found in cases of AIH/PBC overlap. This presumably exemplifies the antigen specificity of type-II AIH in terms of disease pathogenesis. Moreover, most patients with PBC tend to present over the age of 45 whereas anti-LKM-1-positive AIH is most prevalent in childhood and adolescence. Anti-dsDNA antibodies have been suggested as a potential serological marker for AIH/PBC overlap,[73] but this needs further validation.

AIH/PSC-overlap

This is a syndrome with overt cholangiographic or histological findings typical of PSC, alongside robust histological features of AIH concurrently or historically. The development of PSC and AIH in adults usually occurs in a sequential manner, typically presenting with features of AIH first, PSC being diagnosed several years later in some cases.[74, 75] The development of PSC should be always be considered in the AIH patient who is poorly responsive to immunosuppressive therapy; however, as the radiological hallmark of PSC (beading and stricturing of the biliary tree on cholangiography) is a late manifestation of disease, the exact interval between onset of AIH and PSC is difficult to determine. Moreover, cholangiographic features may not be demonstrable even when histological evaluation suggests intrahepatic ductal changes suggestive of PSC. Improvements in future immunohistochemistry techniques during routine liver biopsy may aid the identification of early biliary involvement.

One study evaluated 79 patients given a diagnosis of AIH and found that 10% had MRI findings consistent with a cholangiopathy. Younger age at diagnosis (24.3 ± 11.9), higher baseline ALP (186.4I ± 98.3 IU/L), higher bilirubin at time of MRI (45.8 ± 37.2 mg/dL) and greater lobular activity on initial liver biopsy were significantly associated with detection of this form of AIH/PSC-overlap (P < 0.05), but not ALP/AST ratio, time between the diagnosis of AIH and the MRI or the presence of cirrhosis on initial liver histology.[76] A limitation of MRCP in cirrhosis of any aetiology is that it may demonstrate a biliary tree which simulates a sclerosing pattern, although in this situation beading changes are usually limited to the periphery,[77] common bile duct involvement being restricted to PSC.

Difficulty with case definitions has led to varying estimates as to the exact frequency of the coexistence of AIH and PSC. Figures obtained from the current literature exhibit considerable variation and quote anywhere between 1% and 53.8% of PSC patients as having features of AIH, and ~7–14% of patients with AIH having features of PSC.[75-77] Up to 16% of patients with AIH have coexisting IBD, of which 42% have cholangiographic changes of PSC.[78] The latter is strongly associated with IBD, and the prevalence of IBD in patients with AIH/PSC overlap has been reported to be comparable to that observed with PSC alone. This makes PSC the likely primary disorder in such cases.[60]

Autoimmune sclerosing cholangitis (AiSC)

A form of sclerosing cholangitis associated with florid autoimmune features, including ANA and SMA positivity, high titres of IgG and interface hepatitis has been particularly described in children.[79] This condition, named ‘autoimmune sclerosing cholangitis (AiSC),’ may cause diagnostic confusion as biochemical markers of cholestasis may be normal. Up to 50% of paediatric AIH patients have cholangiographic abnormalities suggestive of sclerosing cholangitis, a quarter of them not having biliary features detectable histologically. Thus, MRCP is advocated for all children diagnosed with AIH. AiSC presents similarly to AIH, although a gender predisposition is not observed. Despite the parenchymal liver damage in AiSC responding well to immunosuppression, biliary disease progresses in about 50% of patients despite therapy.[79]

Applying IAIHG criteria to diagnose AIH overlap in patients with PBC and PSC

The present simplified International Autoimmune Hepatitis Group (IAIHG) criteria[57] facilitate making the diagnosis of pure AIH with a specificity and sensitivity of ~90%.[80-83] Despite the diagnostic accuracy, these criteria and their predecessors need to be applied in the correct context for their utility to be maximised and alternative diseases to be still excluded. Although widely applied to diagnose overlap syndromes in patients with an existing diagnosis of PBC or PSC, the IAIHG scoring criteria was never intended for such use. When applying the IAIHG scoring criteria to a large cohort of patients (n = 479) within the spectrum of autoimmune liver diseases, 7% of PBC and 14% of PSC patients scored as probable or definite AIH, although they did not clinically represent overlap cases.[84] The low sensitivity of the scoring system for defining overlap syndromes in patients with established PBC or PSC is in keeping with results of another study by Papamichalis et al.[85] The IAIHG propose that although patients may have overlapping features across the spectrum of autoimmune liver disease, individual cases should be categorised according to the predominant disease entity, AIH, PBC or PSC; and that the IAIHG scoring system should not be used to diagnose distinct subgroups of patients.[59]

Treatment and Outcome

Treatment decisions must be tailored to the individual, must not be static, and must respond to clinical changes. Guidelines exist purely to provide clinicians with the framework from which to base care. In AIH, immunosuppression in the form of prednisolone combined with azathioprine forms the mainstay of inducing and maintaining remission. Alternative treatments are typically applied to patients either following failure of prednisolone and azathioprine to induce or maintain remission (‘salvage therapies’), or due to the development of side effects/complications (Table 3). The currently accepted definition of complete remission in AIH is a complete normalisation of all inflammatory parameters including AST, ALT and bilirubin (biochemical remission); IgG (immunological remission); and liver histology.[86, 87] This is the goal of therapy in AIH, and achieved in ~75–80%[88, 89] of AIH patients after 24 months with good long-term survival (80% at 10 years). However, the de novo rate of cirrhosis development is 12% after 10 years despite therapy,[90] and this increases the risk of liver-related death/transplantation (HR 9.96) with a reduced 10-year survival (67% vs. 94% in noncirrhotics).[90, 91] Although medical treatment of AIH is very effective, ~10% of patients, require orthotopic liver transplantation (OLT)[92]; the indications being similar to those for other chronic liver diseases. In the small number who present with fulminant hepatic failure, steroid therapy is rarely helpful and liver transplantation may be the only life-saving option.

Table 3. Drugs used in autoimmune hepatitis and overlap syndromes
AgentAdvantageDisadvantageSide effectsComments
  1. a

     Bone protection in those with prolonged steroid use is recommended.[1]

  2. b

     Measuring thiopurine methyltransferase (TPMT) activity has not shown to be useful in predicting azathioprine toxicity in AIH. The role of monitoring 6-thioguanine levels is not completely clear in AIH either.

First line therapy in AIH
Prednisolone

Unequivocal efficacy in AIH

Cheap and widely available

Safe in pregnancy

Clinician familiarity

Associated with a plethora of side effects and complications precluding long-term use

Poor wound healing

Diabetes mellitus

Cosmetic disturbances (acne, weight gain)

Osteoporosisa

Mainstay of inducing remission in AIH

Low doses often required to maintain remission as well

Liver biopsy advised prior to steroid withdrawal

Azathioprine

Unequivocal efficacy in AIH

Cheap and widely available

Safe in pregnancy ( may increase risk of preterm birth)

Clinician familiarity

Steroid-sparing effect proven

Intolerance reported in 10–30%

Associated with developing lymphoma

Frequent blood monitoring needed

Delayed onset of action

Contraindicated in breastfeeding

Leucopenia

Rash

Hepatic dysfunction (rare)

Pancreatitis

Hair loss

Mainstay of maintaining remission in AIH

Compliance may be checked by monitoring thiopurine metabolitesb

Alternative/adjunctive therapy
Budesonide[113-117]

Limited systemic side effects due to first-pass metabolism

Induces remission in 58-83% of AIH patients

May induce remission more rapidly than prednisolone

Expensive

Not suitable for cirrhotic patients

Minimal (usually in those who are cirrhotic)

Increased risk of hepatic and portal vein thrombosis in cirrhosis[118, 119]

Often reserved for young, noncirrhotic AIH patients, or those in whom prednisolone is relatively contraindicated

Anecdotal data supporting its use in AIH/PBC overlap

Further studies needed for validation

Mycophenolate mofetil[120-125]

Quick onset of action

Mostly well tolerated

Teratogenic

Limited experience

Expensive

Diarrhoea

Bone marrow suppression

Pancreatitis

Limited experience based on small case series and open-label studies in AIH with inconsistent results

Benefit mostly confined to those who are azathioprine intolerant rather than azathioprine resistant.

Cycloporine A[126-131]and Tacrolimus[132-134]

Quick onset of action

Safe in pregnancy

Clinician familiarity

Used for induction and maintenance of remission in AIH

Side effects common

Serum levels require monitoring

Need for continued therapy – relapses reported on dose reduction

Hypertension

Renal dysfunction

Hyperlipidaemia

Neurotoxic

Evidence supported by small case series in AIH

Results from the paediatric literature are encouraging

Longer term safety data needed.

Cyclophosphamide[135]

Inhibits antibody production

Quick onset of action

Severe side effects

Toxicity restricts use to experienced centres

Cardiotoxicity

Interstitial pulmonary disease

Pancreatitis

Experience in AIH limited to small uncontrolled series, suggesting that patients may respond well

Used in combination with corticosteroids.

UDCA[136, 137]

Benefit in Japanese patients

Favourably alters HLA expression

May prevent cellular apoptosis

No proven efficacy outside of few isolated reports from the Far East

Does not permit reduction in steroid dosage

Gastrointestinal intolerance: bloating, flatulence

Commonly used in AIH/PBC or AIH/PSC overlap

Benefit outside of PBC remains to be proved

Individuals with overlapping features anecdotally appear to benefit from varying combinations of UDCA and corticosteroids (±azathioprine), but this strategy has little evidence base given the small number of patients comprising few, nonrandomised, nonblinded studies. Moreover, many early reported studies did not use the currently accepted definition of biochemical remission in AIH when evaluating and comparing treatment outcomes; some adopting a fall in ALT or AST <2xULN whereas others refer to treatment response as merely an improvement in liver biochemistry compared with baseline levels. Broadly speaking, treatment should focus on the disease that appears to be the predominant entity.

The limitations of using liver biochemistry as a surrogate of treatment success in overlap must be recognised. For instance, ALT is a good surrogate of disease activity and treatment response in clear-cut AIH, but it is not necessarily the case in overlap presentations and similarly so for ALP. Furthermore, biochemical response to UDCA treatment is validated as prognostic in PBC but is not so in PSC, and extrapolating benefit to patients based on surrogates applied in settings distinct to those from which they were established must be done with caution, and with clear explanation of the limitations to the patient.

AIH/PBC

It is well established that in patients with PBC, UDCA (15 mg/kg/day) leads to slowed progression of fibrosis and liver failure, particularly in patients who demonstrate an adequate biochemical response to therapy.[93, 94] There are several models which have been developed to evaluate UDCA response in PBC, the Corpechot criteria being the most commonly used. In this model, UDCA responders (defined as: AST <2xULN, ALP <3xULN, bilirubin <1 mg/dL after 12 months of therapy) have a 90% 10-year survival (vs. 51% in UDCA nonresponders).[93] However, in patients with florid interface hepatitis on biopsy, the progression of fibrosis may be rapid, and in this situation the institution of immunosuppression may be considered.[66-68]

A retrospective study conducted by Chazouillères et al. compared treatment regimens and outcomes in 17 patients with AIH/PBC overlap over a median of 7.5 years.[67] First-line treatment was UDCA in isolation (n = 11) or combined with corticosteroids (n = 6; 4 also received azathioprine). In the UDCA-alone group, biochemical response and stable/reduced fibrosis was observed in 3/11 patients; the remaining 8/11 were nonresponders and fibrosis worsened in four. In the group receiving both UDCA and immunosuppression, 4/6 patients (all noncirrhotic) achieved a biochemical (transaminases ≤2xULN) and immunological (IgG < 16 g/L) response along with nonprogression of fibrosis. Favourable biochemical responses to combination therapy in AIH/PBC overlap have also been reported by other centres,[71, 95] although the overall frequency of nonresponders is greater than that observed in AIH alone.[96]

A study from the Mayo clinic in 2007 reported that over an average 5.75-year follow-up, more patients with AIH/PBC overlap developed portal hypertension (54% vs. 28%; P < 0.01), features of decompensated disease and progressed to transplantation or death (38% vs. 19%; P < 0.05) compared to patients with AIH alone.[97] However, when Joshi et al. evaluated 16 patients with PBC and overlapping features of AIH, the median change in serum biochemistry and immunoglobulin values was similar to a cohort with isolated PBC after 2 years of UDCA alone. Moreover, very little change in hepatic lobular inflammatory activity was observed during this period.[64] Thus it remains unclear whether the clinical outcome of AIH/PBC overlap is different to that of isolated AIH or PBC. Many studies are limited by the fact that the patients with probable AIH overlap in the context of PBC were less likely to have received UDCA from the outset compared to those with PBC alone. Furthermore, data pertaining to the total duration of UDCA therapy have not been rigorously evaluated.

AIH/PSC

The combination of UDCA and immunosuppressive therapy may improve liver biochemistry in AIH/PSC-overlap syndrome,[74, 75, 98, 99] and this approach has been advocated by international guidelines.[100, 101] It must be emphasised that there are no double-blind, randomised-controlled trials evaluating the efficacy of such a strategy, and although immunosuppressive therapy benefits the hepatitic component of AIH, no survival benefit has conclusively been shown with UDCA in PSC. Treatments in AIH/PSC overlap should thus be individualised based on liver biochemistry, autoantibodies, immunoglobulin titres, cholangiography and histological findings. It is often the presence of a transaminitis >5xULN in a patient with existing PSC, which prompts the search for overlapping AIH, and certainly in the presence of bridging necrosis on liver biopsy it is reasonable for immunosuppressive therapy to be the first line of treatment, UDCA being added once the inflammatory activity has settled. Of note, the study by Boberg et al. demonstrated that PSC patients who have a significant transaminitis were more likely to achieve a biochemical improvement (but not normalisation) with corticosteroids compared to the classical PSC patient with a more cholestatic presentation.[98] However, the early phase of any acute cholestatic process may itself cause a significant elevation of serum aminotransferase levels and in this situation it may be helpful to initiate treatment with UDCA prior to the introduction of any immunosuppressive therapy, as the former improves biochemical indices of cholestasis.

In a prospective Italian study, seven AIH/PSC-overlap patients treated with UDCA, prednisolone and azathioprine demonstrated a significant reduction in serum AST over 5 years (r 2 = 0.7591; P < 0.05), although changes in ALT, ALP and IgG failed to reach significance. No significant changes in liver biochemistry were detectable in the group with ‘classical PSC’ for those who were treated with UDCA alone.[75] Judged from the effect on serum aminotransferases alone, a good response was observed in a Swedish study of 16 AIH/PSC-overlap cases treated with corticosteroids and azathioprine (tacrolimus in one).[99] Of interest, a better biochemical response using immunosuppression in large-duct compared to small-duct PSC was detected, although less than half the patients received UDCA.[99] In 2008, a study from the UK reported that a similar proportion of patients with AIH/PSC overlap and AIH (87.5% vs. 96%; P = N.S.) achieved biochemical and immunological response (normal AST, bilirubin and IgG) within 1 year of therapy, although the exact combination of azathioprine, prednisolone and UDCA varied between individual patients.[96]

Unlike in PBC and AIH, biochemical improvement in PSC does not necessarily translate into better long-term clinical outcome, nor does it predict survival free of liver-related complications or death.[102, 103] Moreover, the majority of patients of European descent with PSC have coexisting IBD, the presence of which has been shown to adversely influence liver-related outcomes.[60] Given the lack of an effective therapy in PSC (the majority developing end-stage liver disease ~17 years following diagnosis[60]), it is unsurprising that individuals with AIH/PSC overlap have a poorer outcome when compared with AIH alone, with more patients failing immunosuppressive therapy,[78, 104] progressing to liver failure or OLT.[76, 105] Patients with AIH/PSC also have a significantly reduced survival compared with AIH alone (HR >2).[96] However, survival is improved compared with patients with isolated PSC.[76, 106]

Conclusions

When managing patients with autoimmune liver disease, a long-term, longitudinal approach to care must be adopted, that focuses on clarity and accuracy in disease definition and rationale for intervention. Overlap presentations are not infrequently encountered and clinicians must adopt a personalised approach to the interventions they offer their patients, ensuring that the predominant disease is managed first, and that patients are well educated about the strength of evidence for interventions offered. Furthermore, clinicians must remain mindful of not over diagnosing overlap presentations, given the inherent spread of presentation of the varying features clinically distinguishable in autoimmune liver disease (and the mimics that can arise from drug injury for example), as well as acknowledge to patients the limitations of surrogates of disease activity when selecting and evaluating the efficacy of treatments. Given the greatest challenge for clinicians being the classification of overlap presentations (realistically because of our limited insights into disease, and the absence of true diagnostic tests for any of the family of autoimmune liver diseases), the long-term goal may lie in the ability to reclassify patients with autoimmune liver disease through discovery of molecular signatures, which more closely capture the disease activity in any given patient. Harnessing the power of genomics and proteomics may succeed in advancement of this lofty long-term goal.

Acknowledgement

Declaration of personal interests: Dr Gideon Hirschfield has served as a consultant for Intercept, Medigene and Janssen Pharmaceuticals. Declaration of funding interests: Dr Palak Trivedi has received research funding from the Wellcome Trust Clinical Research Fellowship Program.

Ancillary