Potential conflict of interest: Nothing to report.
Primary biliary cirrhosis (PBC) is a chronic cholestatic liver disorder that can progress to cirrhosis, shortening life expectancy. PBC patients are often asymptomatic, present with biochemical cholestasis, and test positive (≥90%) for antimitochondrial antibodies (AMAs) in serum. Although AMA positivity without biochemical cholestasis may indicate increased risk of future PBC development, the contribution of these antibodies to pathogenesis remains enigmatic. Environmental risks and genetic determinants are likely implicated in PBC etiology. Given the familial aggregation of PBC, we hypothesized that AMAs also aggregate among relatives of PBC probands. We investigated the prevalence of AMAs in first-degree relatives (FDRs) of PBC probands to examine whether AMAs aggregate in such pedigrees. Using a PBC family registry, we prospectively screened for AMAs in the serum of 306 FDRs in 145 pedigrees, 350 PBC probands, and 196 controls who were age-matched, sex-matched, race-matched, and residence-matched to probands. The prevalence of AMA in FDRs and controls was 13.1% and 1%, respectively. Greater prevalence of AMA was found in female FDRs of PBC probands [sisters (20.7%), mothers (15.1%), and daughters (9.8%)] than in male FDRs [brothers (7.8%), fathers (3.7%), and sons (0%)]. Conclusions: AMAs aggregate among FDRs of PBC probands. Our data have clinical implications for FDRs of PBC probands because AMA positivity may suggest susceptibility to PBC. Thus, the identification and follow-up of these relatives may lead to earlier disease diagnosis and treatment. Furthermore, if AMA development is heritable, this trait will provide a basis to dissect the genetic predisposition to PBC. (HEPATOLOGY 2007.)
Primary biliary cirrhosis (PBC) is a progressive cholestatic liver disease primarily affecting women and characterized by immune-mediated destruction of the intrahepatic bile ducts, gradually leading to cholestasis and cirrhosis.1 Interestingly, about 50% of PBC patients have no symptoms and are incidentally diagnosed following abnormal results in routine liver tests.2 The early detection of PBC is important because timely treatment with ursodeoxycholic acid before the development of late-stage disease appears to normalize life expectancy.3–5
The pathogenesis of PBC is considered to be complex, the result of interplay between environmental exposure and unknown genetic susceptibility alleles. Nongenetic risk factors associated with PBC thus far include past smoking,6, 7 a history of urinary tract infection,7 and hormone replacement therapy.7 In addition, a number of microorganisms, such as Novosphingobium aromaticivorans, Chlamydia pneumoniae, and betaretrovirus, have been proposed to have a role in PBC development.8–10 The genetic contribution to PBC pathogenesis is also thought to be important, as evidenced by high disease concordance in monozygotic twins11 and increased familial risk for this disease.12, 13
To date, little progress has been made toward identifying the putative genetic factors of PBC pathogenesis. This stems from the complexity of the currently postulated multihit disease model in which PBC initiation (first hit) and consequent progression to clinical disease (subsequent hits) represent distinct disease processes,14, 15 each potentially evolving from a spectrum of genetic and environmental interactions. Furthermore, a lack of sufficiently large DNA collections from PBC patients, family members, and matched controls has hampered past efforts at identifying genetic risk factors because of limited statistical power.
The established serological hallmark of PBC is the presence of antimitochondrial antibodies (AMAs) reactive to the E2 subunits of 2-oxodehydrogenase enzymes, most specifically to those of the pyruvate dehydrogenase complex (PDC-E2).16 These antibodies are present in the serum of approximately 90% of patients with PBC.1 Indeed, AMA positivity is virtually diagnostic of PBC when coupled with a persistent biochemical cholestasis.17 Moreover, the detection of AMAs in an otherwise healthy individual without abnormal liver tests may indicate a risk for the future development of overt disease.18, 19
The major epitope of PDC-E2 recognized by AMAs overlaps with those of autoreactive B- and T-cells shown to specifically target the biliary epithelial cells lining the intrahepatic bile ducts of PBC patients.17 The process by which self PDC-E2 moieties become antigenic in PBC has been widely studied, and several mechanisms, including molecular mimicry, self alteration by xenobiotics, and intact immunogenic epitopes released from apoptotic biliary epithelial cells, have been suggested.20 Conceivably, any of these mechanisms could play a role in the initiation of the autoimmune response, which is likely dependent on the individual's genetic makeup and history of environmental exposure. The presence of AMAs in serum before the development of clinically recognizable PBC can be viewed as a marker of an early disease process, namely, the loss of immune tolerance to select PDC-E2 epitopes. However, an understanding of the underlying mechanism(s) linking AMA development to clinical PBC remains elusive. The dissection of the factors involved in AMA development in PBC families may prove to be a useful strategy for assessing early PBC pathogenesis and setting the stage for examining the progression to clinical disease. Therefore, determining the prevalence and putative inheritance of AMAs in families of PBC probands will provide a framework for such future inquiry.
A handful of previous studies examining the prevalence of AMAs in PBC families have been performed, but they have generally been limited in scope. In the largest of these studies, Feizi et al.21 tested 118 first-degree relatives (FDRs) of 27 PBC probands for AMAs by immunofluorescence and found that 7.6% (9 of 118) had detectable antibodies in serum. However, almost half of the FDRs in that study were children of the probands, with a mean age of 26 years. Galbraith et al.22 tested for AMAs in 95 biological relatives (mean age of 47 years, relationship unspecified) of 22 PBC probands and reported detectable antibodies in 7.4% of the screened relatives. In contrast, Caldwell et al.23 assessed 41 FDRs of 27 PBC probands and found 4.9% to be positive for AMAs. That study included only 16 siblings (sex not specified) and many children of the probands. In a study from Italy, Floreani et al.24 tested for AMAs in 115 blood relatives of unspecified relationship to 30 PBC probands and found 6.1% tested positive for the antibody. Finally, in a study from Brazil involving 100 FDRs of 26 PBC probands, only 1 relative was found to have detectable AMAs.25
In the aforementioned studies, the number of relatives screened for AMAs was generally small, the relationship to the proband was often not specified, and many of the tested relatives were either children or young siblings of the PBC probands. These factors limited the appraisal of aggregation of AMAs in PBC families. Our study evaluated the prevalence of AMAs among FDRs of PBC probands with a large, prospective, and comprehensive registry consisting of well-characterized pedigree structures.
Subjects and Development of the Registry and Biospecimen Repository.
We currently enroll PBC patients (i.e., probands), their FDRs (i.e., parents, siblings, and children), and controls matched to the cases into our ongoing Mayo Clinic PBC Genetic Epidemiology (MCPGE) Registry and Biospecimen Repository with the aim of elucidating the genetic and environmental contributors to PBC pathogenesis. The registry and study have been approved by the Mayo Clinic Institutional Review Board.
The enrollment process begins with the identification of PBC probands, primarily those who were evaluated at the Mayo Clinic in Rochester, MN, between January 1994 and December 2005 and also a small number of referred PBC patients. Between November 2002 and June 2006, we mailed study invitation letters to 1348 PBC patients, which explained the study goals and procedures. To date, 746 patients have responded, 531 (71.2%) of whom have agreed to participate, becoming the probands of the study (Fig. 1). The study participants are asked to (i) provide informed consent, (ii) complete the validated study questionnaire (which collects information on personal and family medical history, lifestyle, and environmental exposures), (iii) release contact information for their FDRs through the family information form (FIF), and (iv) submit a sample of blood for (1) liver enzyme [i.e., alkaline phosphatase (ALP) and alanine aminotransferase (ALT)] and AMA testing in serum, (2) the isolation of genomic DNA, (3) the creation of Epstein-Barr virus–transformed lymphoblastoid cell lines, and (4) the storage of serum, plasma, and buffy coat samples for use in future investigations. PBC probands are between the ages of 18 and 90 years at the time of recruitment and fulfill the established disease criteria for PBC.2 Specifically, the diagnosis of PBC is made on the basis of a persistent biochemical profile of cholestasis (greater than 6 months) in the absence of another known cause of liver disease, detectable AMAs in serum, and compatible liver histopathology. The diagnosis of AMA-negative PBC patients is made on the basis of chronic biochemical cholestasis and compatible liver biopsies in the absence of detectable AMAs in serum.
Following the completion of proband enrollment, we invite by separate letter every living FDR (age 18-90 years), using information provided by the proband in the FIF. Between January 2004 and June 2006, we mailed invitation letters to 1368 FDRs of enrolled PBC probands explaining the study. Thus far, 711 have responded, 649 (91.3%) of whom have agreed to participate in the study (Fig. 1). The FDRs are asked to provide the same items as the probands, except for the FIF.
Additionally, we have recruited unrelated controls matched to each PBC proband by age (±2.5 years), sex, race, and state of residence. All controls are recruited from the Mayo Clinic Division of General Internal Medicine during annual visits for routine medical checkups. To date, 298 potential controls have been approached, and 206 (69.1%) have agreed to participate (Fig. 1). The controls are asked to provide the same items as the PBC probands, except for the FIF. We offer no compensation to any participant in the MCPGE registry.
Collection of Biospecimens.
The collection of blood specimens from PBC probands, FDRs, and matched controls is facilitated by bar-coded, mail-in kits, which are prepared and distributed by the Mayo Central Laboratory for Clinical Trials. The blood specimens are either collected on-site at 1 of the outpatient phlebotomy stations at the Mayo Clinic in Rochester, MN, or collected off-site at the convenience of our study participants and shipped to us by overnight courier following blood draw. The blood processing and packaging instructions, assessment of the received blood samples, and quality control are performed under the supervision of the Mayo Central Laboratory for Clinical Trials.
Clinical Assessment and Biochemical Evaluation of the Participants.
We review the available medical records of recruited PBC probands and matched controls. Questionnaires, FIFs, and pedigrees of the study participants are evaluated for accuracy, and data are extracted as indicated.
Serum from MCPGE registry participants is prospectively tested for serum ALP, ALT, and AMAs during the study enrollment process. Biochemical testing for ALP and ALT is performed in the Mayo Clinic General Biochemistry Laboratory according to the instructions of the manufacturer (Roche/Hitachi Diagnostics, Indianapolis, IN). The AMA testing is performed by the Mayo Clinic Diagnostic Immunology Laboratory with a commercially available enzyme-linked immunosorbent assay (ELISA; Diastat AMA, Euro-Diagnostica, Malmo, Sweden) specific for the PDC-E2 (M2) antigen. The resulting absorbance units are classified as negative (≤0.1 units) or positive (≥0.2 units) for AMA detection.
MCPGE Registry and Biospecimen Repository.
The number of active participants in the MCPGE registry is currently 508 PBC probands, 648 FDRs, and 196 individually matched controls. Of these, we have received completed questionnaires and probands' FIFs from 379 PBC probands, 501 FDRs, and 154 controls. The biospecimen repository presently includes blood-derived specimens from 357 PBC probands, 310 FDRs, and 196 matched controls.
AMA Test Results in PBC Probands, FDRs, and Matched Controls.
To date, we have assessed AMAs in 350 PBC probands, 306 FDRs, and 196 matched controls, and the results are shown in Table 1. The probands are primarily female (91.7%), with a mean age at PBC diagnosis of 51.0 years, which is consistent with published epidemiological data.7 As expected, AMA was detected in nearly 90% of the probands. The demographics and AMA data of the 350 PBC probands were essentially the same as those of the subgroup of the probands with FDRs tested for AMAs (n = 145; data not shown).
Table 1. Sex, Mean Age, and Percentage of AMA-Positives Among PBC Probands, Matched Controls, and FDRs of PBC Probands
The group of 306 FDRs consisted of 80 parents, 162 siblings, and 64 children from 145 PBC pedigrees. These FDRs constitute 32.5% of the living relatives of the probands, representing 69.4%, 36.6%, and 16.6% of the living parents, siblings, and children, respectively. The highest prevalence of AMAs occurred in the females of each relative category, with 15.1% of mothers, 20.7% of sisters, and 9.8% of daughters having detectable AMAs (≥0.2 units). In contrast, AMA prevalence in fathers, brothers, and sons was 3.7%, 7.8%, and 0%, respectively. AMA was detected at low levels (i.e., 0.2 units) in 1.0% (2 of 196) of the controls. The difference in the proportions of AMA-positive FDRs (13.1%) and controls (1%) was statistically significant (P < 0.001 by χ2 testing). The mean age of the AMA-positive and AMA-negative PBC FDRs did not differ by relationship group, with the exception of sons, because none of this category had AMAs in serum (Supplementary Fig. 1). Interestingly, the cumulative AMA positivity in PBC FDRs by age revealed a sharp increase in the rate of AMA positivity among female FDRs at age 35 that continued to increase gradually with advancing age. Of interest, the male PBC FDRs had no AMA positivity until age 60 without any notable increase in the rate of AMA positivity beyond that age (Supplementary Fig. 1).
In a separate retrospective analysis of medical records, we found that 85.3% of the first 217 consecutive PBC probands in this registry tested positive in the past for AMA by immunohistochemistry (titer > 1:40). To this end, the present ELISA testing for AMAs at the cutoff level of 0.2 units demonstrated that 88% of the PBC probands of this registry had detectable antibodies in serum. Additionally, prospective AMA testing of our primary sclerosing cholangitis biospecimen repository, which includes 127 patients with an established diagnosis of primary sclerosing cholangitis, revealed that none of these participants tested positive (≥0.2 units) for AMAs with the same ELISA method.
AMAs and Familial History of PBC.
The familial structure and history of PBC extracted from the questionnaires of all 379 participating PBC probands were compared between the subset of 145 PBC probands in which the proband and at least 1 of the FDRs had been tested for AMAs in the study and the subset of 234 probands without FDRs tested for AMAs (Table 2). The percentage of female patients and age of PBC diagnosis were similar between these 2 groups. However, we did detect a bias toward multiple-affected families in the subset of probands with FDRs tested for AMAs. Among these 145 probands, 12.4% reported having 1 or more FDRs with PBC versus 6.8% of the 234 probands without FDRs tested for AMAs and 9.0% for the overall sample. Moreover, the reported prevalence of PBC in the FDRs of the subset of 145 probands with FDRs tested for AMAs was also slightly higher than that of the other 234 probands (1.7% versus 1.1%). However, neither observation was significant by χ2 testing. Further analysis shows that the response bias present in our subset of FDRs tested for AMAs is due to the overrepresentation of sisters known to have PBC, a fact not surprising given the aims of our study.
We also analyzed the data after removing FDRs who were known to have PBC by proband or relative report, thereby conservatively addressing the effect of response bias. This resulted in a decrease of the overall AMA prevalence from 13.1%-10.2% among the PBC FDRs, most notably because of mothers (15.1%-12.2%), brothers (7.8%-6.0%), and sisters (20.7%-14.9%; Tables 1 and 3). To further understand the extent and effect of this bias on our data, we considered the familial history of PBC (proband plus ≥ 1 FDR with PBC) and the reported PBC status of the FDRs (Table 3). A total of 39 FDRs of the 18 probands reporting a familial history of PBC had been tested for AMAs. Of these, 11 were known to have PBC (by proband questionnaire response), 10 (90.9%) of whom were found to be AMA-positive. Of interest, the AMA-negative FDR in this group is the sister of an AMA-negative PBC proband. The remaining 28 FDRs of the familial PBC probands had not been previously diagnosed with PBC, and 3 (10.7%) were found to be positive for AMAs. Of these, 2 were brothers of AMA-positive female PBC probands who had reported having sisters with PBC (1 of these sisters was tested by us and also found to be AMA-positive), and 1 was the sister of an AMA-negative PBC proband who reported another sister to have PBC (who was tested and also found to be AMA-positive). Among the 257 FDRs of 127 probands who reported no familial history of PBC, 26 (10.1%) had detectable AMAs, including 13.3% of mothers, 15.6% of sisters, and 12.5% of daughters.
Table 3. AMA Results in FDRs Not Known to Have PBC or a Family History of PBC
History of Familial PBC
No History of Familial PBC
Number of families
Relationship to the proband
Results of Serum Liver Tests in AMA-Positive FDRs.
Only 1 of the 29 AMA-positive FDRs not previously known to have PBC had a significant elevation of ALP, which is suggestive of disease. This individual did not report a history of PBC in the study questionnaire. Among the remaining 28 FDRs, there was either no or mild elevation of ALP or ALT. None of the 29 AMA-positive FDRs have undergone liver biopsy for PBC evaluation. The follow-up of these FDRs will provide an opportunity to understand the natural history and clinical importance of AMA involvement in disease pathogenesis.
Using a large family registry, we have demonstrated that AMAs aggregate in FDRs of PBC probands, emphasizing the involvement of familial factors (be they genetic and/or shared-environment) in AMA development. Evidence for a substantial genetic component of PBC includes high disease concordance among monozygotic twins11 and increased familial risk of the disease.12, 13 Moreover, a recent epidemiologic study has shown that the greatest individual risk factor for developing PBC is having an FDR with the disease.7 Nevertheless, genetic studies in PBC are hampered by the rarity of the disease and potentially by diagnosis bias, as the disease is often asymptomatic for years. AMAs are a strong indicator of PBC when accompanied by persistent biochemical cholestasis. AMAs may also be predictive of future PBC development. For example, the presence of AMAs has been shown to foreshadow the development of PBC signs and symptoms in the majority of individuals so affected.18, 26 However, these patients did not represent the healthy population but were tested for AMAs as a result of rheumatologic and thyroid complaints.26 To date, no study has been able to address the mechanisms of AMA development as a risk factor or predisposing agent to clinical PBC in humans. Better evidence for AMA development before PBC comes from a recently reported mouse model that captures several features of PBC pathogenesis.27 Indeed, these NOD.c3c4 congenic mice develop detectable AMAs to PDC-E2 before biliary damage.27
Given the strong familial component of PBC risk and the conjecture that AMAs may be predictive of PBC, we have hypothesized that AMA development is a heritable trait aggregating among relatives of PBC probands and may be related to the increased familial risk of PBC. To this extent, AMA status could prove useful in studies aimed at shedding light on the genetic and environmental factors implicit to the familial predisposition to PBC and subsequently to those involved with PBC pathogenesis in general. To begin testing this hypothesis, we have investigated the prevalence of AMAs in FDRs of PBC probands, using our PBC family registry. We report the prevalence of AMAs in FDRs of PBC probands to be 13.1%, which is much higher than that seen in controls (1%). The highest prevalence of AMAs was found in sisters (20.7%), mothers (15.1%), and daughters (9.8%) of PBC probands. Even after the exclusion of those FDRs known by report to have PBC, the prevalence of AMAs was still high (10.2%) independently of the PBC family history, including 14.9% of sisters, 12.2% of mothers, and 10.5% of daughters. Our observation that 3.7% of fathers and 7.8% of brothers have detectable AMAs was surprising because of the historically low prevalence of PBC in males. This might suggest the presence of sex-specific protective mechanisms (i.e., genetic or environmental) that preclude these relatives from developing PBC despite manifesting a putatively predisposing factor to the disease (i.e., AMAs). Moreover, although in the general population PBC is 10 times less common in males than in females, the PBC epidemiology study by Gershwin et al.7 illustrates that the ratio of female-male PBC prevalence is less than this 10-fold difference in FDRs of PBC patients (mothers, 1.7%, to fathers, 0.5%, a 3.4-fold difference, and sisters, 4.3%, to brothers, 0.8%, a 5.4-fold difference). These data suggest that the PBC family history presents an additional burden on the disease risk to males, as further evidenced by our finding regarding AMAs in male FDRs of PBC patients.
Earlier studies conducted to evaluate the prevalence of AMAs in PBC families have been limited.21–25 In 2 of these investigations, almost half of the FDRs were children of the probands, so their age may have lowered the estimate of the prevalence of AMAs among FDRs.21, 23 Moreover, in the studies by Floreani et al.24 and Galbraith et al.,22 the relationship of the family members to PBC probands was not specified (i.e., first-degree or second-degree). Lack of complete data on the familial structure from these studies compromises any direct correlation to our study. Although our present study is the largest familial assessment of AMAs in FDRs to date, we realize its limitations, including the cross-sectional design. We plan to continue to extend this cohort by enrolling additional FDRs and to construct a replicate sample in the future.
In our study, we found the prevalence of AMAs among matched controls to be 1% (2 of 196). Moreover, both of the AMA-positive controls demonstrated the lowest level of the antibody (0.2 units) at the cutoff to be considered positive in our assay. Others have reported the prevalence of AMAs in normal populations to be as high as 0.6%.28, 29 Nevertheless, these studies did not assess the prevalence of AMAs in healthy controls matched for age and sex to a cohort of PBC patients. In another study, the prevalence of AMAs among corporate workers undergoing annual health check-ups was 0.44% (5 out of 1145); however, when the data were stratified by sex, the AMA prevalence in females was 0.91%, and all 5 of the affected individuals were females older than 40 years of age.30 This observation underscores the significance of sex and age in interpreting the results of AMA prevalence in normal populations. To this end, our study is the largest known collection of controls who were age-matched, sex-matched, race-matched, and residence-matched to PBC patients with prospective testing for AMAs performed to date.
Our observation that AMAs aggregate with a relatively high prevalence among FDRs of PBC probands has direct clinical significance and important research implications. The diagnosis of PBC at its earliest stages and prompt treatment with ursodeoxycholic acid can normalize patient survival.3–5 Thus, screening for AMAs in asymptomatic middle-aged female FDRs of PBC probands is reasonable. Individuals who were positive for AMAs could then be regularly tested for liver enzymes, and if they were elevated, a liver biopsy would be warranted to assess for PBC. Although previous reports suggest that PBC will develop in many AMA-positive nonbiologically related individuals,18, 26 it is currently unclear what proportion of asymptomatic AMA-positive PBC FDRs will develop disease. The complexity of shared genetic and environmental factors in these families could lead to increased risk of AMA development unrelated to those factors necessary for the advancement to full-blown clinical PBC. Ideally, a large prospective study should be pursued to better determine the predictive value of AMAs on PBC development in PBC FDRs and controls. Such a study would lead to a greater understanding of PBC pathogenesis by allowing the observation of the early disease process as it unfolds.
The significance of AMAs to PBC pathogenesis remains mysterious 40 years after their discovery as a diagnostic marker of PBC.31 By focusing on the aggregation of AMAs in PBC families, this study introduces a novel approach to dissecting the puzzle of PBC pathogenesis. Familial aggregation of AMAs generally implies that AMA development lies in the spectrum between complete genetic inheritance and the sole effect of a shared environment. Current evidence suggests a substantial level of genetic influence on the development of PBC, which likely overlaps with presence of AMAs. A formal segregation analysis of AMA development in PBC pedigrees will help to verify the existence of a genetic influence and begin to define an inheritance model of this trait. These results will place us in the unique position of developing a linkage analysis to study the genomic loci that contribute to AMA development in PBC families. Such an inquiry will unquestionably lead to a better understanding of the pathogenesis of PBC.
We are indebted to the patients with PBC, their family members, and the healthy controls who participated in this study. We thank Stacy Roberson for excellent secretarial assistance.