Epidemiology of autoimmune liver disease


Dr J Heathcote, Toronto Western Hospital, 6B Fell Pavillion, 399 Bathurst Street, Toronto, Ontario M5T 2S8, Canada. Email: jenny.heathcote@utoronto.ca


Primary biliary cirrhosis (PBC), autoimmune hepatitis (AIH) and primary sclerosing cholangitis (PSC) are chronic liver diseases that likely have an autoimmune basis to their pathogenesis. Although significant strides have been made in the clinical management of these conditions, their pathogenesis remains obscure. Understanding of various epidemiological factors may shed light on predisposing or causative factors for these diseases. Most is known about the epidemiology of PBC, with only minimal information on that of PSC and AIH. In this review, the current data on the epidemiology of PBC, AIH and PSC are summarized and suggestions are made for future work in this important area.


Primary biliary cirrhosis (PBC), autoimmune hepatitis (AIH) and primary sclerosing cholangitis (PSC) are chronic liver diseases that likely have an autoimmune basis to their pathogenesis.1–3 While these diseases are generally distinguishable based on clinical grounds, occasionally patients will exhibit features suggestive of an overlap syndrome.4 Unfortunately, the etiology of all three diseases remains obscure. Epidemiology can be a powerful tool by which to gain important insights about diseases based on their associations. Over the last number of years, more sophisticated techniques have been applied to better understand the epidemiology of autoimmune liver disease, however, many questions remain unanswered. In this review, we summarize what is currently known about the epidemiology of PBC, AIH and PSC and make some suggestions for future work in this important area.


The vast majority of studies examining the incidence and prevalence of autoimmune liver disease have focused on PBC. Estimates of both disease incidence and prevalence vary quite widely between studies of specific defined populations ( Table 1). The highest annual incidence and point prevalence rates for PBC have been found in the UK (incidence 3.1 per 100 000 population per year; prevalence 25.1 per 100 000 population)19 and the USA (incidence 2.7 per 100 000 population per year, prevalence, 40.2/100 000).18 Estimates from Sweden and Norway (incidence 1.4–1.6/100 000 per year and prevalence 9.2–15.1/100 000) are also fairly high when compared with those reported outside of northern Europe.6,8,11,16 The disease appears to be rare in Asia, with an annual incidence of 0.34–0.42 per 100 000 population and a prevalence estimated at 2.0 per 100 000 population in Japan.14 Similarly, only five cases have been reported from India.20 Interestingly, Canada and Australia have found significantly lower incidence and prevalence rates (0.33/100 000 per year and 1.9–2.2/100 000 population, respectively) than those reported in England, despite having strong British heritages in both countries.10,12 Some of this disparity likely reflects true regional variation; however, study design almost certainly accounts for a portion of the observed differences.

Table 1.    Studies of incidence and prevalence of primary biliary cirrhosis
Study and reference no.YearCases (no.)Annual incidence/100 000Prevalence/100 000Methods
  1. AMA, follow up of all positive tests for antimitochondrial antibodies; EASL, European Association for the Study of the Liver; HR, review of hospital records; NA, data not available; PS, physician survey.

Sheffield, UK51980    34NA  5.4PS, AMA
Malmo, Sweden61984    331.4  9.2HR
EASL71984  5690.4  0.5–7.5PS
Orebro, Sweden81985    361.412.8AMA
Newcastle, UK91989  3471.1–1.915.4PS, AMA, HR
Ontario, Canada101990  2060.33  2.2PS
Umea, Sweden111990    861.315.1PS, AMA, HR
Victoria, Australia121993    84NA  1.9PS, AMA, HR
Estonia131995    690.23  2.7PS, HR
Japan14199512580.34–0.42  2.0HR, AMA
Newcastle, UK151997  1602.218–24PS, AMA, HR
Oslo, Norway161998    211.614.6HR
Newcastle, UK171999  7703.125.1PS, HR, AMA
Olmstead County, USA182000    462.740.2PS, AMA, HR

To aid in standardization and accuracy of future epidemiological studies of PBC, Metcalf and James proposed a set of specific guidelines to be followed.21 These guidelines are (i) stringent case inclusion criteria; (ii) clear definition of date of disease onset; (iii) well-defined study period, area and population; (iv) multiple case-finding methods; and (v) rigorous tracing of all possible cases.21 Illustrative of the importance of these strategies is the observation that 23.4% of patients in a study in north-east England were detected through a case-finding mechanism that included tracing all patients with positive antimitochondrial antibody testing, reviewing hospital admission records, as well as surveying all general internists in the region. It was also noted that up to 37% of patients in this study had not previously been identified as having PBC, despite having symptoms, signs or laboratory abnormalities consistent with this diagnosis.19 Together these observations suggest that a significant number of cases of PBC may be missed if studies rely on physician surveys only.

Given the marked heterogeneity in study design, it is difficult to be certain of the importance of regional variation in disease incidence and prevalence and it is noteworthy that the most rigorous studies have identified the highest rates of PBC, despite coming from different populations.18,19 Regional variation is an important issue to clarify, as environmental and genetic factors have been proposed as being relevant to disease pathogenesis.22 In an early study from Triger in Sheffield England, the prevalence of PBC in residents using a particular water reservoir was 10-fold that of those using other reservoirs.5 Despite thorough investigation of the water source, no cause for this apparent disease clustering was identified.5 Recently, Prince et al. used statistical methods to demonstrate the presence of non-random spatial variations of PBC prevalence in a well-defined population in northern England.7 While the observations suggest a possible environmental cause or trigger for the disease, no such agent has yet been identified.

In addition to possible regional variation, it has been observed that rates of PBC may be increasing with time. Between 1987–1994, prevalence rates in the UK have increased significantly from 14.9 to 25.1 cases per 100 000 population (P < 0.00001). Over the same period, there has been some variation in annual incidence with a general, non-significant, increasing trend (2.3/100 000 in 1987 to 3.2/100 000 in 1994).19,23 These figures may represent true increases in disease; however, other explanations must also be considered. More exhaustive case-finding strategies and generally increased awareness of PBC likely account for some of the rise in incidence, and earlier diagnosis could increase prevalence without altering incidence. It will be important to examine PBC rates over time in other populations to see if this is an isolated finding in the UK or a more generalized phenomenon.

Circulating antimitochondrial antibodies (AMA) have long been considered the hallmark of PBC.1 They are present in 95% of patients with the disease and longitudinal follow up has shown that the majority of asymptomatic AMA-positive patients will develop PBC.24 However, despite this observation, prevalence studies have shown that AMA positivity is much more common than PBC itself.25–27 Part of this apparent disparity may be related to the natural history of the disease. Because PBC is generally very slow to progress, a proportion of AMA-positive patients will remain asymptomatic and therefore unrecognized, possibly lifelong. Although such individuals likely have histological changes compatible with PBC, they will not be identified in most studies of PBC prevalence. However, further comparisons of AMA and PBC prevalence are needed to better understand this apparent paradox.

Unfortunately, considerably less is known about the incidence and prevalence of AIH and PSC. Much of the information regarding AIH comes from studies of idiopathic chronic active hepatitis (CAH). Prior to the introduction of the Autoimmune Hepatitis Group Revised Scoring System, there was no standardized way of evaluating patients with suspected AIH and, as a result, the studies included somewhat heterogeneous patient populations.28,29 Only one study has examined the epidemiology of AIH directly. In a Norwegian population, Boberg et al. found a mean annual incidence of 1.9 cases of AIH per 100 000 and a point prevalence of 16.9 cases per 100 000 population.16 Although they did not use the AIH scoring system for diagnosis, this was the first and remains the only study of patients with AIH in whom hepatitis C infection was excluded. Using similar case identification methods, they also calculated the incidence and prevalence of PBC and PSC in the same population. They found that the incidence and prevalence of AIH were similar to that of PBC (incidence 1.6/100 000 and prevalence 14.6/100 000) and almost double that of PSC (incidence 1.31/100 000 and prevalence 8.5/100 000). The rates of AIH found in this study were about twice those found in studies of idiopathic CAH in Iceland30 and AIH as a cause of CAH in Sweden31 and England.32

Type II autoimmune hepatitis, associated with liver–kidney–microsomal (LKM) antibodies, is much less common than type I AIH; however, no formal studies of incidence or prevalence rates have been performed. It was first formally described in 1987 and appears to be much more common in southern Europe than in either northern Europe or the USA.33,34 Clearly more study is needed to better define the global prevalence of this important variant of AIH.

Because of the marked heterogeneity of patient populations and the inability to exclude hepatitis C in the earlier studies of AIH, it is difficult and likely inappropriate to compare study results. A rigorous epidemiological study of AIH applying the AIH revised scoring system for diagnosis is needed to accurately define the burden of this disease.

Aside from the Norwegian study that compared rates of PBC, AIH and PSC, the only other population-based epidemiological study of PSC comes from Spain.35 Escorsell et al. demonstrated an increasing annual incidence and prevalence of PSC from 1985 to 1988. In 1988 they found the incidence of PSC to be 0.068/100 000 with a point prevalence of 0.22/100 000 population. Other estimates of incidence and prevalence of PSC are extrapolated from ulcerative colitis (UC) data. Assuming that 3–5% of patients with UC have concomitant PSC, the prevalence of PSC in the USA has been estimated to be 1–4/100 000.3 This compares with a similarly generated estimated value of 6/100 000 prevalence in Sweden.36 Given the strong association of PSC with UC, it is likely that PSC is rare in regions with a low prevalence of inflammatory bowel disease (IBD).


Despite sharing a common autoimmune pathogenesis, the three autoimmune liver diseases are characterized by fairly distinct demographic patterns of disease. Primary biliary cirrhosis is most commonly found in women, typically between 40 and 65 years of age. The youngest reported patient with PBC was 15 years old at the time of diagnosis.37 There is a marked female predominance with a female to male ratio (F/M) ranging from 6:1 to 22:1 across various studies.7,13,38,39 Some studies of incidence and prevalence of PBC have calculated rates for specific age and sex cohorts. In a study from the Mayo clinic, Kim et al.18 found that the mean annual incidence (1975–1995) of PBC per 100 000 person years ranged from 0.7 for men to 4.5 for women. Similarly, in the UK, the incidence of PBC in the whole population was 32.2/100 000 compared with 43.0/100 000 in adults over age 20 years and 100.4/100 000 in women over age 40 years.19 Although much less common in men, PBC seems to follow a similar clinical course in male patients.40

Autoimmune hepatitis was originally described as a disease of young women; however, more recent reports have identified a broad age range among patients. A biphasic distribution for age at presentation has been noted with peaks between 10–30 years and then >40 years of age.41 However, a recent large study of AIH found similar proportions of patients presenting in each decade of life (Feld et al., unpubl. data, 2002). Type II AIH occurs almost exclusively in women and tends to have a younger age of onset.42

In contrast to AIH and PBC, PSC is much more common in men, with a F/M ratio ranging from 0.2 to 0.4.3,43 Patients usually present between 20–40 years of age;44 however, PSC is being increasingly recognized as a cause of chronic liver disease in children. The two largest series have documented over 80 children with cholangiographic evidence of PSC, including 15 infants.45–47

Very little is known about variation of autoimmune liver disease by ethnicity. Most studies reported in English have come from Europe and North America and have included exclusively or primarily Caucasian patients. A recent study by Lim et al. showed that African Americans with AIH are more likely than Caucasians to present with cirrhosis.47 They also tended to require higher doses of immunosuppression to achieve remission than Caucasian patients; however, the long-term outcome was similar in both groups.47 Data on all three diseases from Japan and India indicate that although present, the diseases are relatively uncommon in Asia. 48–51 As globalization continues it is likely that future reports in high-prevalence countries will be better able to report the ethnic breakdown of subjects. If environmental triggers play a role in disease pathogenesis, one should be able to observe a change in disease prevalence in immigrant populations.


Genetic factors have been shown to play an important role in many aspects of a variety of autoimmune diseases. In addition to conferring susceptibility or providing protection from a specific disease, they also may modify disease onset, progression and/or severity. This has led investigators to examine the role of genetic predisposition in autoimmune liver disease by looking both at familial disease and specific genetic markers.

All studies of familial PBC have documented a markedly increased prevalence of the disease among first-degree relatives of patients compared with the general population. Bach and Schaffner performed a retrospective chart review and prospective mail survey to assess the prevalence of PBC in the relatives of 405 family members. They identified 26 patients with PBC who had at least one family member affected by the disease, resulting in a prevalence of 6.4% among first-degree relatives.52 This figure is greater than 150-fold higher than the highest reported population prevalence (40.2 per 100 000 population).18,53 In a British study, Brind et al. found 10 of 736 patients with PBC had a positive family history, leading to a familial prevalence of PBC of 1.33%.54 Ten of 156 patients with PBC in Hiroshima, Japan, were found to have a family member with the disease, reflecting a familial prevalence of 5.1% compared with only 5.4 cases per 100 000 (0.0054%) in the general population.55 Similarly, a prospective study of 156 patients with PBC in Italy found six (3.8%) definite cases of PBC among relatives.56

The majority of studies looking at familial risk of PBC have examined case series in specific liver units as opposed to geographically defined populations. The observation that this study design has the potential to overestimate familial risk led Jones et al. to examine familial PBC in Newcastle-upon-Tyne in the UK, where rigorous epidemiological studies have defined the prevalence of PBC in the community.19,23 They identified 10 families among 160 patients (6.4%) with at least two first-degree relatives with PBC. These families were quite large (mean number of first-degree relatives 7.1), resulting in an absolute disease prevalence in first-degree relatives of 0.72%. Based on the prevalence of PBC in this community, the relative risk for first-degree relatives compared with the general population can be estimated at 18.4 with a sibling relative risk of 10.5 and an offspring relative risk of 30.6. If probable cases of PBC are included, the relative risks rise to 34.4 for first-degree relatives, 20.9 for siblings and 45.4 for offspring. Given that all estimates do not account for relatives who will develop PBC in the future, or those who have asymptomatic, undiagnosed disease, these are minimum values. While this study found a slightly lower familial risk than previous studies, it confirms a significant risk for PBC among family members of patients compared with that of the general population.

The majority of familial cases of PBC have been identified in mother–daughter pairs and interestingly, in all cases, daughters presented at a younger age than their mothers. Although disease awareness may have led to earlier diagnosis, none of the daughters were diagnosed because of screening instigated by maternal disease. This pattern of earlier disease in successive generations may suggest genetic anticipation within families in PBC.53,56 The only reported case of PBC in a monozygotic twin found the presence of the disease to be discordant; however, the monozygosity in this case was not unequivocally established.57,58

Combined, this evidence is suggestive that there is a genetic component to PBC; however, other factors likely also play a role. The absence of PBC in childhood, the marked female predilection and the relative rarity of familial disease, make it unlikely that PBC is a simple genetic disease.53

Familial risk of AIH has not been rigorously studied. There are few reports of AIH occurring in family members and no twin studies.59–62 While AIH itself does not appear to be common in relatives of patients, autoimmune disease per se is quite prevalent. In fact, it is such a common finding that the presence of autoimmune disease in a first-degree relative is included in the Autoimmune Hepatitis Group Revised Scoring System for AIH diagnosis.28

A number of individual cases of familial PSC have been reported; however, no systematic study has examined familial risk.63 The largest series included just three cases of siblings with PSC, all of whom also had UC.64 Familial clustering of IBD has long been recognized and the recent identification of the NOD2 gene increasing susceptibility to Crohn's disease strongly supports a genetic component to IBD.65,66 The strong association between PSC and IBD, particularly UC, makes genetic susceptibility an attractive hypothesis for PSC; however, to date there is little pedigree data to support this notion.


The majority of work looking for specific genetic markers predisposing to autoimmune liver disease has focused on the coding region for the major histocompatibility complex (MHC) on chromosome 6. All three diseases have been studied extensively for human leukocyte antigen (HLA) associations ( Table 2).

Table 2.    Selected human leukocyte antigen (HLA) associations with autoimmune liver diseases
HLA MoleculePrimary biliary cirrhosisAutoimmune hepatitisPrimary sclerosing cholangitisComment
  1. –, Data not available; NA, no association found.

DR8 DRB1*08IncreasedCaucasian patients67–73
DR3Increased69,74Aggressive disease41,75,76Increased77Associations independent of other DR3-associated autoimmune diseases
DPB1*0301Increased78Caucasian patients
DPB1*0501Increased79Japanese patients
DRB1*0803Increased79Japanese patients
DR4NAIncreased (in DR3-negative patients)41,76,80,81Protective (when present may predict aggressive disease)82Commonest AIH association in Japan.80 Mild/later onset AIH.41 If present in PSC, may predict aggressive disease, but possibly protective for disease development.82
DRB4*0103Concurrent autoimmune disease76 
DRB1*0301Poor treatment response77 
DRB1*0401Mild disease77 
DRB1*1301/DQB1*0603Argentinian/Brazilian children83,84Adults from same region associated with DR4 alleles84
DRw52aIncreasedInitially reported 100% concordance85
DR2Increased (DR3 neg patients)Children and adults47,77
DR3/DR2Aggressive disease77,86 

A large number of studies have examined the association of MHC class I molecules with PBC, documenting no association with the disease.68,74,87–89 In contrast, associations with a variety of MHC class II molecules have been documented. The most common finding has been the identification of HLA-DR8 (DRB1*08) as a risk factor for PBC;67–73 however, others have also noted DR362,89 and DPB1*030178 in association with the disease. Among Japanese patients, associations with DR2,87 DPB1*0501,79 and DRB1*080390 have been identified with PBC as well. None of these associations has been found unanimously and the overall association of MHC class II with PBC is clearly not as strong as that with a number of other autoimmune diseases.91,92 This has led a number of investigators to look at other genetic associations with PBC including polymorphisms in the tumor necrosis factor-alpha promoter region,93,94 the TAP1/TAP2 genes on chromosome 6,95 the promoter region of the interleukin-10 gene,96 the natural resistance associated macrophage protein 197 gene and the vitamin D receptor gene.98 Further studies of familial PBC may help clarify the genetic components of this disease.

As with PBC, many investigators have looked at HLA associations with AIH. The most commonly identified HLA associations were DR3 and DR4; however, patient heterogeneity and failure to adequately exclude viral infection hampered early linkage studies. Donaldson et al.41 examined a fairly homogeneous population of 96 northern European Caucasian patients and found that HLA A1-B8-DR3 as well as HLA-DR4 were independently associated with type I AIH. They also noted that the A1-B8-DR3 haplotype occurred in younger patients who were more likely to relapse off therapy, while DR4 was associated with older patients with seemingly milder disease. Later studies confirmed these disease associations and their respective patterns of disease,75,76 leading to the inclusion of HLA DR3/DR4 as diagnostic criteria in the Autoimmune Hepatitis Group Revised Scoring system.28 Subsequently, other patterns of disease have been identified  with  HLA  typing.  The  presence  of  DRB4*0103 is associated with concurrent autoimmune disease, DRB1*0301 with poor response to corticosteroid therapy and DRB1*0401 with milder disease.76

Most of the initial HLA studies of type 1 AIH were conducted in American and European Caucasian patients. When other patient populations have been examined, some interesting patterns have emerged. In Japanese patients with AIH, the A1-B8-DR3 haplotype is very uncommon and disease susceptibility is almost exclusively associated with DR4.80,81 A novel and strong HLA association (HLA-DRB1*1301-DQB1*0603 haplotype) was identified in Brazilian and Argentinean children with AIH,83 whereas adults from these countries were more likely to have DR4 alleles.84 Human leukocyte antigen-DR has also been found to be associated with type 2 AIH, however, the specificities of the DR region differed from those with type 1 disease.42,99 Examining the associations found in different populations may provide a unique opportunity to discriminate relevant genes within a tightly linked genomic region.100

There was great initial excitement with the report by Prochazka et al. of 29 adult patients with PSC, all of whom possessed the DRw52a antigen.85 However, while some subsequent studies have found increased rates of this antigen among patients with PSC, the 100% concordance has not been confirmed.101,102 As with AIH and PBC, DR3 has also been found in association with PSC independent of the presence of other autoimmune diseases known to be DR3-associated.103 DR2 is also found in patients with PSC, but primarily in DR3-negative patients.46,77 As with AIH, certain alleles have clinical significance. Mehal et al. found that the presence of HLA-DR4 was a marker for rapid disease progression in patients with PSC;82 however, others have not confirmed this pattern.104 Recently Boberg et al. reported that in a large series of European patients, those who were DR3/DR2 heterozygotes had an increased risk of death or liver transplant.86 Clearly, further research in this area is needed to clarify these clinical associations, given their potential prognostic significance. In order to be confident in the strength of any associations found it will be important to ensure future studies are of adequate sample size, as many past investigations examined only small cohorts of patients.


Similarly to many other autoimmune conditions, autoimmune liver disease is associated with a variety of other illnesses thought to have an autoimmune pathogenesis. A fairly extensive list of immune-mediated diseases has been reported in association with PBC, AIH and PSC ( Table 3). The most striking association is found between PSC and IBD. Approximately 70% of adult patients with PSC have concomitant IBD at some stage in their lives, while only about 5% of patients with IBD develop PSC.125 Primary sclerosing cholangitis and IBD are reported to be less commonly associated outside of Western Europe and North America,126–128 as well as in children with PSC.45,46 Most patients present with IBD prior to the onset of PSC; however, in some series up to 50% have been diagnosed with liver disease before any evidence of colitis was documented.46 Primary sclerosing cholangitis occurs with both Crohn's disease and UC, but it is much more commonly found with UC and colonic involvement is universal in patients with PSC and Crohn's disease.3

Table 3.    Disease associations of autoimmune liver diseases
Primary biliary cirrhosisAutoimmune hepatitisPrimary sclerosing cholangitis
Keratoconjunctivitis sicca105Autoimmune thyroiditis106Ulcerative colitis3
Xerostomia105Grave's disease107Crohn's disease108
Sjögren's syndrome105Ulcerative colitis109Histiocytosis X110
Scleroderma/CREST syndrome111Autoimmune hemolytic anemia112Rheumatoid arthritis113
Rheumatoid arthritis114Idiopathic thrombocytopenia115Colorectal cancer116
Autoimmune thyroiditis117Systemic lupus erythematosus118Cholangiocarcinoma3
Celiac disease119Sjögren's syndrome117Celiac disease120
Mixed connective tissue disease114Polymyositis114Retroperitoneal fibrosis121
Renal tubular acidosis122Mixed connective tissue disease114 
Celiac disease123 
Myasthenia gravis124 

Given the strong association between PSC and UC, epidemiological patterns in UC have been examined in PSC as well. Two factors that have been clearly shown to alter the risk of UC are smoking behavior and previous appendectomy. Ulcerative colitis is much less common among smokers,129,130 and three studies have documented a negative association between previous appendectomy and risk of UC.131–133 Van Erpecum et al.134 looked at these two factors and their relation to PSC and found that although PSC was more common among non-smokers independent of the presence of UC, no association between PSC and previous appendectomy was identified. Long-standing UC is an important risk factor for the development of colorectal cancer (CRC). Kornfeld et al. found that PSC was also a risk factor for the development of CRC independent of the presence of UC.116 Based on this, it has been recommended that patients with PSC be placed in a colonoscopic screening program regardless of the activity or presence of colitis.125

In addition to the associations with autoimmune disease, PBC has also been reported to be associated with cancer. Early reports indicated an increased risk of extrahepatic cancer in general and a 4.4-fold increase in the standardized incidence ratio (SIR) for breast cancer in particular.135–137 In contrast, later studies from Italy, Holland, Sweden and Denmark found no association between PBC and breast cancer.138–141 Most recently, Howel et al.142 examined a large, well-characterized population of patients with PBC in northern England and found a small but statistically significant excess of cancer incidence at all sites in patients with PBC (SIR = 1.7) but no increased risk of breast cancer (SIR = 1.1). They also found a markedly increased risk of hepatocellular carcinoma (HCC) among patients with PBC (SIR = 74.3). While some other studies have also found increased numbers of HCC among patients with PBC,143–145 a large Swedish study by Loof et al.140 found an SIR of only 2.9 for HCC. The increased risk of HCC likely relates to the presence of established cirrhosis and the inclusion of patients with predominantly early-stage disease in the Swedish study may have masked this result. Although there may be a very small increase in all cancers in patients with PBC, the risk appears to be mostly attributable to the presence of HCC in cirrhotic patients.


In the majority of patients with autoimmune liver disease, a specific diagnosis can be made with confidence. However, in a proportion of patients, features of more than one disease may be present. So-called ‘overlap syndromes’ have been described with all combinations of PBC, AIH and PSC, except for the combination of PBC and PSC.

The only large study of these variant forms of autoimmune liver disease comes from the Mayo clinic, where 162 patients with AIH, 37 with PBC and 26 with PSC were evaluated.4 Despite their initial clinical diagnoses, they were found to have features of another disease in 12%, 19% and 54% of patients, respectively. While these figures suggest that the ‘overlap syndromes’ are very common, a significant referral bias must be considered when interpreting the results of this study. It would be valuable to look at a well-defined community-based population to determine the true incidence and prevalence of these diseases in order to better estimate their significance in clinical practice. Clear, universally accepted definitions of the different overlap syndromes are also required.

The most common overlap appears to be between PSC and AIH. This is particularly true among children. In a Canadian series, nine of 32 patients with PSC had clinical features entirely in keeping with AIH, including a compatible biopsy in five patients. In addition, 17 patients had an elevated serum immunoglobulin G level, while 19 had circulating antismooth muscle antibodies and 14 were antinuclear antibody positive. The diagnosis of PSC was confirmed with cholangiography in all cases.46 Similar findings have been documented in other smaller series.146,147 Most recently, Gregorio et al. looked at a large series of English children with clinical AIH and found that 27 of 55 (49%) had cholangiographically proven PSC, suggesting that this overlap syndrome is much more common than previously recognized.48 Based on this observation, cholangiography is recommended in children with AIH or idiopathic chronic liver disease, particularly if they have a progressive clinical course or do not respond well to corticosteroid therapy.46

In an attempt to define the predominant disease in patients with overlap syndromes, Czaja et al. recently looked at HLA associations. Unfortunately, despite confirming the known HLA disease associations with PBC, AIH and PSC, they did not find that HLA type helped predict the clinical course of overlap patients.148


Primary biliary cirrhosis, AIH and PSC form the spectrum of autoimmune liver disease. Although distinct, they share many common characteristics. Great strides have been made in the understanding of the epidemiology of these diseases; however, further rigorous epidemiological studies are needed to truly define the incidence, prevalence, familial risk and disease associations of PBC, AIH and PSC in a variety of populations over time. This information will lead to advances in the understanding of the definition and the pathogenesis of these complex diseases.