Autoimmune disease overlaps and the liver: Two for the price of one?


Correspondence: DrIrMackay Department of Biochemistry and Molecular Biology, Monash University, Wellington Road, Clayton, VIC 3168, Australia. Email:

See article on page 95

Autoimmune diseases aggregate in the one patient and/or family members: such clustering was an early defining marker for an autoimmune disease. 1 Examples are provided by the clustering of autoimmune thyroiditis with gastritis and insulitis (diabetes), Sjögren’s disease with rheumatoid arthritis, primary biliary cirrhosis (PBC) with the CREST syndrome, and autoimmune hepatitis (AIH) with haemolytic blood diseases, thyroiditis, ulcerative colitis and others. 2 These clustered diseases are independent entities, but there is an underlying pathogenetic determinant in common: this is a defect in the establishment and maintenance of self tolerance, that is, autoimmunity and/or an aberrant inflammatory response. The independent existence within a cluster of two or even more autoimmune or inflammatory diseases poses less of a problem than when there is coexpression of two usually distinct disease entities within a single organ, in which case the term ‘overlap’ is used. One early example was the overlap of Hashimoto’s thyroiditis and thyrotoxicosis, dubbed ‘Hashitoxicosis’. In the case of the liver, several such coexistences or overlaps are recognized, and there is much uncertainty as to how they should be defined, named and classified. As background, Table 1 summarizes features of the known or suspected autoimmune diseases of the liver.

Table 1.  Autoimmune diseases affecting the liver
DiseasePopulation biasBiochemistryHistologySerologyHLAOverlaps
  1. AIH, autoimmune hepatitis; PBC, primary biliary cirrhosis; AIC, autoimmune cholangitis; PSC, primary sclerosing cholangitis; CAH, chronic active hepatitis; LFT, liver function test; Ig, immunoglobulin; HLA, human leucocyte antigen; ANA, antinuclear antibody; n.e., not established; anti-LKM, antibody to liver–kidney microsomes; AMA, antimitochondrial antibody; pANCA, perinuclear type of anti-neutrophil cytoplasmic antibody; SMA, smooth muscle antibody. Information was assembled from Krawitt et al.3 and Manns. 26

AIH type 1Northern europid;Hepatitic LFT, markedInterface and lobularANA, homogeneousB8, DR3PBC
 female 6:1increase IgGhepatitis; plasma cellsSMA, F-actinDRB1*0301AIC
   (= CAH) DR4PSC
AIH type 2Middle europid;Hepatitic LFT, markedInterface and lobularAnti-LKM, (anti-CYP2D6)n.e.Nil
 female; childhoodincrease IgGhepatitis; plasma cells 
   (= CAH) 
AIH, secondaryn.e.Hepatitic LFTCAHAs for AIH type 1 orn.e.Hepatitis C
    type 2; titres lower Hepatitis B
      Wilsonian CAH
      Alcoholic CAH
PBCNorthern europid;Obstructive LFT;Biliary ductular damage;AMA (anti-2-OADCDR8AIH
 female 10:1increase IgMgranulomasenzymes)DRB1*0801AIC
    ANA in ~ 50%(Europids)
AICFemaleObstructive LFT;Biliary ductular damage;ANA (centromere, Sp-100,n.e.AIH
  increase IgMgranulomasnuclear pore complex) PBC
PSCEuropid; male 2:1Obstructive LFTBile duct lesions;pANCAB8, DR3AIH (in children)
   periductular fibrosis DRB1*0301 


Historically, in 1969, there was a viewpoint that chronic active hepatitis (the predecessor of autoimmune hepatitis of today) and PBC could be linked in the unified concept of autoimmune liver disease. 4 An early study revealed that in 11% of cases of chronic active hepatitis (CAH) there was a positive test for antimitochondrial antibody (AMA) together with clinical and histological features of both CAH and PBC, 5 and MacSween, in a personal experience covering both diseases, noted that histopathologIcal differentiation was difficult in some 15% of cases. 6 The concept of ‘overlap’ or ‘mixed form’ disease, with features of both CAH and PBC, was formally introduced in 1977 by Klöppel et al.7 In 1980, the CAH–PBC overlap syndrome was given a serological basis, together with yet another descriptive term of chronic cholestatic hepatitis by Berg et al. who reported 17 cases of which all were positive for auto-antibodies usually associated with CAH, and also for the AMA typical of PBC, together with a new antimitochondrial reactant designated M4. 8 In 1992, Davis et al. reported on the CAH–PBC overlap syndrome, citing 11 cases that were clinically typical, although not studied for the auto-antibodies characteristic of AIH: their cases were seropositive for anti-M2 but an anti-M4 type of reactivity was not demonstrable. The report of the International Autoimmune Hepatitis Group (IAIHG), 10 convened in Brighton in 1992 to generate diagnostic criteria for AIH, stated that the overlap of AIH and PBC was well recognized, but such cases comprised only a ‘small proportion’ of the total case-load in most centres and, hence, elected to defer preparation of diagnostic criteria.

Intrafamilial occurrence of single or multiple autoimmune diseases is well recognized. Among the liver diseases, this applies more to PBC 11 than to AIH despite the weaker human leucocyte antigen (HLA) associations in PBC than AIH. Hence, it is surprising that reports are so scarce on the intrafamilial overlap of AIH with PBC. There is one published example, that of two elderly sisters of whom one had PBC and the other had AIH with concurrent myxoedema, with various tissue auto-antibodies being demonstrable among the relatives. 12

The perceived frequency of overlap may well depend on the standpoint of the observer, whether from a primary interest in AIH or PBC. From an AIH standpoint, Kenny et al. described 187 Mayo Clinic patients with severe, chronic, active (presumably autoimmune) hepatitis among whom they identified 37 (20%) with a positive test for AMA. 13 These 37 patients also had higher serum levels of alkaline phosphatase (ALP) and an increased frequency in liver biopsies of stainable hepatic copper (features suggestive of PBC), but 91% of the 37 cases could be allocated to a CAH category; only four (2.1%) of the 37 had titres of AMA > 1/40, and only two (1.1%) had histological features typical of PBC. From a PBC standpoint, Chazouillères et al. assessed clinical features and response to therapy in 12 patients (10 females, two males, median age 50 years), selected from 130 consecutive patients with a primary diagnosis of PBC, and for whom diagnostic criteria for both diseases were fulfilled. 14 For 11, there were histological features of both diseases at the time of clinical presentation and, for one, an initial presentation of ‘pure’ PBC was followed by a typical flare of AIH. The authors concluded that the frequency of the AIH–PBC overlap syndrome among patients with PBC was 9%, there was usually a coexistence ab initio of both diseases, and that treatment should be directed to both diseases (i.e. a combination of ursodeoxycholic acid (UDCA) and a corticosteroid drug). In a further Mayo Clinic study, Czaja analysed overlap or ‘variant’ syndromes based on 223 patients with primary diagnoses of AIH (162), PBC (37) or primary sclerosing cholangitis (PSC; 26). 15 Of the 162 cases with AIH, there were eight (5%) with coexisting PBC as judged simply by a positive test for AMA, and of 37 with PBC there were seven (19%) with coexisting AIH as judged by the IAIHG scoring system. 10 Detailed histological or immunoserological comparisons for overlap cases in the various groups were not given in any of the studies cited above.

The second report from the IAIHG, prepared in 1999, discussed the overlap of AIH with various cholestatic syndromes, and concluded that ‘all of these disparate disorders (if, indeed, they are distinct) are still in need of a universally agreed definition and classification’, and suggested that advice be obtained from an international working party. 16 Meanwhile overlaps could be dealt with either by nominating the independent occurrence of two diseases, given that criteria for both are fulfilled, or by defining criteria for each overlap syndrome as such. The latter seems preferable, given a likely shared immunopathological basis for both conditions, and the known tendency for escalation among autoimmune diseases. More detailed serological investigation should help. Apart from the conventional auto-antibodies described above, diagnostic pointers may be provided by the perinuclear type of anti-neutrophil cytoplasmic antibody (pANCA) which occurs frequently in AIH 16,17 but seldom in PBC, and perhaps also by other AIH-related auto-antibodies (e.g. to soluble liver antigen (anti-SLA), or to liver cytosol antigen type 1 (anti-LC1). 18


The entity known as autoimmune cholangitis/cholangiopathy (AIC) has itself been thought to represent an overlap disease. This was first described in 1987 as a chronic obstructive intrahepatic inflammatory cholangitis in which the biochemical features were those of PBC, but the serological features differed in that the test for AMA was negative and that for antinuclear antibody (ANA) was positive, and there was a response to corticosteroid drugs. 19 This entity subsequently attracted a number of case studies and commentaries, recently summarized and reviewed in a conjoint study on 24 cases of AIC from Japanese colleagues and this laboratory. 20 This study compared clinical, biochemical, histological and serological features of AIC with those of AIH and PBC. The results refuted the idea that AIC is a cholangiopathic variant of AIH and indicated that so-called AMA-negative PBC 21 may be too simplistic, yet were consistent with the current viewpoint that AIC aligns closely with PBC, clinically and histologically. 20 Also, in AIC, the expressions of ANA by immunofluorescence and immunoblot are those that occur in the 50% of ANA-positive cases of PBC, namely auto-antibodies to centromere, the Sp-100 antigen, and anti-nuclear pore proteins including gp210, its aminoterminal fragment and/or p62 nucleoporin. 20 Studies by immunoblot on our 24 cases of AIC disclosed that, whereas AMA was negative by immunofluorescence, there was often serological reactivity demonstrable by immunoblot with components of the M2 autoantigen family, the E2 subunits of the 2-oxo-acid dehydrogenase complex (2-OADC) enzymes: this is in accord with positivity by immunoblot in cases of AMA-negative PBC. 22 Thus, we concluded that AIC is part of PBC but, by reason of the distinctive serological features, it should be regarded as a nosological entity, at least until the actual pathogenesis of PBC itself is revealed. 20

The above having been said, might AIC in fact coexist with AIH? Some reports indicate this to be the case. Ben-Ari et al. recorded the shared histological features of periportal and introlobular inflammation and bile duct damage, and biochemical remission with prednisolone, 23 and Czaja stated that 11 of 162 cases (17%) of AIH had features of AIC. 15 Li and co-authors, in this issue of the Journal, report the case of a 42-year-old Chinese female who expressed ‘characteristic features of both diseases’. 24 The biochemical features included highly raised serum levels of ALP and γ-glutamyl transpeptidase, increased serum cholesterol and normal immunoglobulin (Ig)G but an increased IgM: all are more suggestive of PBC or AIC than of AIH. Serologically, AMA was negative but immunoblot to confirm non-reactivity to 2-OADC-E2 enzymes was not performed. The test for smooth muscle antibody (SMA) was negative and ANA was positive, but the specificity of the ANA was that of anticentromere, which is one of the particular trio of ANA reactivities (see above) that mark ANA-positive cases of PBC. Histologically, there was an AIH-PBC overlap, but the presence of periductular granulomas points more particularly to PBC. There was an initial response to prednisolone, and prednisolone dependence, in addition to benefit from UDCA. Thus, the case of Li et al.24 shows sufficient features to justify being styled an overlap, but the bias seems much more towards AIC than AIH.


Autoimmune hepatitis–primary sclerosing cholangitis is an interesting overlap because PSC can be questioned as a ‘qualified’ autoimmune disease.

Taking data from a recent review, 25 the sex bias is towards men (70%); the usual accompanying disease, ulcerative colitis, which coexists in some 75% of cases, is itself questionable as an autoimmune expression; the characteristic histopathology with biliary periductular ‘onion-skin’ fibrosis is not typical for an autoimmune process; there is neither a disease-specific nor tissue-specific auto-antigen; there is a muted, if any, response to corticosteroids or immunosuppressive drugs; and, finally, the miscellaneous immune aberrations that are described 25,26 cannot be readily assembled into a concept of autoimmune pathogenesis. The evidence for autoimmunity that does exist for PSC comprises a high frequency of pANCA 27 and an association with HLA alleles that includes the autoimmune haplotype HLA B8,DR3, and DRw52a. 28 Thus, PSC may well be viewed as immunologically mediated but is, like its partner disease ulcerative colitis, the outcome of an aberrant immune response directed primarily to intestinal luminal antigens, presumably of bacterial origin.

The frequency of overlap of AIH and PSC is hard to estimate. Roberts et al. from the Mayo Clinic reported eight patients (four males, four females) fulfilling criteria for both diseases, with five developing PSC after the clinical onset of AIH, seven having an accompanying ulcerative colitis and all giving a poorer response to corticosteroid therapy. 29 The analysis of overlaps by Czaja included 26 cases of PSC, with a positive IAIHG score for AIH in as many as 14 (54%) of the cases; 15 this may reflect the recognized low specificity of the scoring system in the setting of cholestatic syndromes. 16 Boberg et al. reported on 114 adults with PSC among whom criteria for AIH were met by 40, two definite and 38 probable; 30 however, using a revised scoring system, only 10 of the 38 retained a score for probable AIH. 16 Although cases with the primary diagnosis of AIH seldom show concordant features of PSC, an obstructive pattern of tests of liver function may develop in the later stages of AIH. McNair et al., in 1998, commented on the scarcity of overlap between AIH and PSC in their report of five examples, claiming that of cases reported in the literature, only 11 were sufficiently well documented for comparison with their five cases. 31 Of these five, four were male (a PSC bias), four had a marked elevation of ALP (a PSC bias), all had hyperglobulinaemia (an AIH bias), and two had ulcerative colitis. Immunoserological tests (performed in the hospital laboratory) revealed that ANA was positive in two (titres 1:2560 and 1:320, pattern not specified), SMA in four (an AIH bias) and pANCA in all five. Radiography showed strictured ducts with beading (a PSC bias); histology showed changes of pericholangitis in three cases; and there was a response to prednisolone in all cases (an AIH bias).

While the frequency of overlap of AIH and PSC is low in adults, this may not pertain in children, according to the description by Wilschanski et al.32 although this overlap was not emphasized in the report by Gregorio et al. on 47 cases of AIH in children. 33 Roberts commented on the difficulties in distinguishing between the two conditions on clinical, histological or even serological data and called for further studies to elucidate whether the overlap represented a single disease process with combined features of AIH and PSC, or the concurrence of two ‘mechanistically distinct’ but clinically similar entities. 34


There can be apparent overlaps between AIH and other diseases within the immunoinflammatory spectrum of histological CAH. First, in reference to chronic viral hepatitis, an autoimmune overlay appears to be much more frequent in infection with hepatitis C than hepatitis B. In fact, in chronic hepatitis C infection, there are various immune aberrations including the mixed cryoglobulinaemia syndrome which is due to deposition of immune complexes that include components of HCV, as well as autoimmune serological expressions similar to those seen in either type 1, or type 2, AIH. These are probably provoked by virus infection in the liver. 35 Wilsonian CAH may resemble the necroinflammatory features of AIH histologically 36 but, according to the literature review of Milkiewicz et al. (submitted for publication, 1999), autoimmune serological expressions are rare. Similarly, CAH in alcoholic patients is more a histological than a serological diagnosis. 37 Perhaps in these various examples, autoimmunization is consequential to hepatocellular damage from a primary cause but serves as an additional pathogenetic component.


There may be insights to be gained from immune-mediated liver diseases that do not overlap. One example is type 2 AIH, for which reports of overlap with other diseases are lacking, even with the type 1 variant. Another claimed example of non-overlap among inflammatory liver diseases is PBC and PSC. 26


There remain substantial gaps in knowledge on how autoimmune diseases are initiated and sustained: a provisional outline is presented in Table 2. Consideration of overlap syndromes has conceptional connotations for at least three areas.

Table 2.  Origin of multiple autoimmune syndromes
  1. * HLA (human leucocyte antigen) contributes 20–60% genetic risk; background genes may number approximately 20, even approximately 100. Infection is the most likely trigger, but cause and effect are seldom evident; inflammatory milieu disrupts tolerogenesis; mimicry popular but unproven. As judged by variability in time of onset and disease severity, in autoimmune-prone strains of mice, including NOD. § For review (see Vanderlugt and Miller 39); explains multiple autoimmune reactants. Idea is applicable to double autoimmune syndromes affecting liver.

Autoimmume induction
Permissive geneticsHLA, background genes*
Disease initiation
Infection Inflammatory milieu
  Molecular mimicry
  Virus + host → neoantigen
  Superantigen effect
Other Chemicals
  Drug + host → neoantigen
Stochastic (random) events related to immune
Epitope spreading§
Intramolecular:further epitopes on original auto-antigen
Intermolecular:further epitopes on other auto-antigens
Autoimmume escalation
Secondary, tertiary autoimmune disease expressions
including overlaps during evolution of an autoimmune

First, for nosologists, overlap syndromes pose uncertainty whether there should be separate nomination of the two diseases contributing to the overlap, or whether the overlap should be defined as a single entity. If the latter were the case, nomenclature groups will need to develop appropriate criteria, equivalent to those established or in de facto use for AIH, PBC and PSC. A recognized handicap is that a defining serological marker (or markers) for the common type 1 AIH, equivalent to the anti-CY-P450-2D6 that characterizes type 2 AIH, is still not yet available. For the most-discussed of the overlaps, AIH and PBC, it seems that PBC is the more dominant of the partners onto which features of AIH become added, as judged from observations that this overlap is more frequent from the PBC perspective than the AIH perspective. 13,14

Second, for therapists, medication appropriate to each partner of the overlap should be prescribed. This applies particularly to AIH–PBC and AIH–PSC overlaps, and to AIC, in which a combination of immunosuppression (prednisolone, azathioprine) and UDCA would be logical, and preliminary indications of benefit are already available. 24,38

Third, for theorists, the study of overlaps may be informative for an understanding of the general pathogenesis of autoimmune liver diseases (see Table 2). The recognized genetic elements in autoimmune disease are estimated to comprise 30–60% of the overall risk and, of this, genes of the major histocompatibility complex (AIH) are the main contributor. These HLA alleles, and genes that influence the specificity of B and T cell antigen receptors, determine the direction of responses to specific auto-antigenic molecules and, thus, confer disease individuality. Other immune-response genes will have a general influence, either by effects on tolerogenesis overall, or on apoptosis, on the balance of cytokines produced by T helper -l or -2 cells, or on expression of other inflammatory mediators. The combined effects of such general immune response genes would explain autoimmune escalation in which there is cumulative organ or tissue involvement to produce a complex multisystem disease phenotype. The process of epitope-spreading is described as either intramolecular, which implies spread to different epitopes on the one auto-antigen, or intermolecular, which implies spread to different auto-antigens in the one cell or tissue. 39 This process would be facilitated by a permissive genotype and an inflammatory milieu and ambient cytokines, and also could contribute to disease overlaps and diversification and serological responses to multiple auto-antigens. In the context of genetics, Lohse et al. suggest, as indicated above, that the PBC–AIH overlap syndrome is, in fact, primarily PBC with hepatic expressions conferred by HLA susceptibility alleles B8, DR3 or DR4. 40

In conclusion, overlap syndromes, are a reality among the hepatic autoimmune diseases. These are a nuisance for nosologists, but careful study of clinical, laboratory and particularly the serological features of these syndromes and their outcome, should provide interesting insights for autoimmunity. Moreover, their recognition should prompt the use of novel combined therapies. A clarion call can, therefore, be made for increased international collaborations among hepatologists, including exchanges of histological material, serum samples and DNA for immunogenetic analysis, as no single clinic would have sufficient cases or access to sufficient technological resources, to provide a solution to these ‘double disease’ occurrences. The outcome would be an overlap of benefits for both hepatology and clinical immunology.


The author thanks Dr Ian McFarlane and Dr Senga Whittingham for their helpful suggestions