Background

Primary biliary cirrhosis (PBC) is a cholestatic disease of unknown aetiology primarily affecting middle-aged women. It is characterized by progressive destruction of the small intrahepatic bile ducts that leads to ductopenia, fibrosis and ultimately liver cirrhosis. The serologic hallmark of PBC in 95-98% of the patients are the M2 anti-mitochondrial autoantibodies (AMA), directed against the E2 sub-unit of the pyruvate dehydrogenase multi-enzyme complex (PDC) located in the inner membrane of the mitochondria. [1] It is considered an autoimmune disease resulting from the interaction of multiple environmental factors, the immune system and the liver of genetically susceptible individuals [2].

The role of the genetic factors in PBC is strongly supported by the 63% concordance rate in monozygotic twins, the second highest reported in autoimmunity after celiac disease [3]; as well as the high familial PBC occurrence rates reported 4% [4] to 9% [5] in first-degree relatives (FDR) of PBC patients [6]. Moreover the presence of AMA with no other clinical evidence of disease is reported in 13.1% [5]. The significant coexistence of other autoimmune diseases and/or other autoantibodies in both patients and their FDR, also indicates the genetic background of PBC [7].

Geographic clusters of the disease have been found in small areas, a fact that underlines the significance of environmental factors in the induction of PBC [8–11]. A number of microorganisms (E. Coli [12], Mycobacteria [13], Novosphingovium aromaticivorans [14, 15], Lactobacillus species [16], Chlamydia pneumoniae [17, 18], Helicobacter pylori [19]), viral infections (human beta retrovirus [20], MMTV [21]), xenobiotics [10, 22, 23] and halogenated compounds found in hair dyes, nail polish and cigarette smoke [24, 25] have been associated with PBC as environmental factors.

The age at first pregnancy, the frequency of pregnancies, abortions, hormone replacement therapy (HRT) and a previous history of obstetric cholestasis, as well as urinary tract infections (UTI), tonsillectomy, appendectomy and cholecystectomy have also been reported as possible risk factors [4, 24–26].

Many studies include patients with ethnic variations that may influence the results. Studies in populations that share the same genetic background, common environment and low migration rates make it more plausible to identify environmental and/or genetic factors playing a role in a disease’s induction.

Therefore we analyzed the demographics, lifestyle, medical and surgical history in a genetically and ethnically homogeneous group of Cretan PBC patients and their FDR compared with a control group matched for age, gender and residence with the patients in order to identify familial occurrence rates, associated lifestyle factors and/or comorbidities.

Methods

Between March and October 2007 we mailed study invitation letters to 196 PBC patients and their FDR (parents and children), who were regularly followed at the Department of Gastroenterology and Hepatology of the University Hospital of Heraklion (Crete, Greece), referral centre for liver disease in the island. The letter explained in detail the scientific data concerning the genetic and environmental factors in the pathogenesis of the disease, and pointed out the goals and the procedures of the study for both patients and their FDR. A hundred-eleven patients consented (56.6%), 40 patients declined and 45 patients did not respond. A hundred-fifteen FDR (75 females, 40 males) also consented to be enrolled (55% of the living FDR). A hundred-forty-nine unrelated controls matched to PBC patient by age (±2.5 years), gender, Cretan origin and residence were also enrolled at the study. The unrelated control group was enrolled among the visitors of the hospital.

All study participants, after signing an informed consent, completed a questionnaire through an interview performed by the same doctor. The questionnaire included information regarding demographics (age, gender, place of origin and residence), socioeconomics (profession, educational status), lifestyle (body mass index (BMI) grouped as ‘high’ (>30), ‘medium’ (25–30), and ‘low’(<25), smoking status in 20≤, >20 packs/year including past smoking, weekly alcohol consumption analyzed as “no use” (no consumption), “use” (less than 14 units of alcohol per week for men and 7 units women) or “misuse” (more than 14 units of alcohol per week for men and 7 units for women) counting 1 unit equal to 12gr of alcohol. ‘Hair dye use’ is referred as at least once a month per year, whereas ‘no hair dye use’ as never used. ‘N;ail polish use’ is referred as ≥10 times per year, whereas ‘no nail polish use’ is referred as occasional use or never used. Medical and surgical history questions referred at the period prior to PBC diagnosis. Moreover detailed questions concerning the frequency of vaginal and UTI, thyroid gland dysfunction, chronic diseases (hypertension (HT), diabetes mellitus (DM), dyslipidaemia, coronary artery disease (CAD), peripheral vasculopathy, asthma/chronic obstructive pulmonary disease (COPD)), allergies, other possible liver diseases and other autoimmune diseases were recorded. Diagnosis of Hashimoto’s disease was established by the combination of hypothyroidism and elevated thyroperoxidase (TPO) and thyroglobulin (TG) antibodies. For the female participants a detailed reproductive history prior to PBC diagnosis was assessed, which included the date of menarche, date of first pregnancy, number of pregnancies, childbirths, abortions or miscarriages; menopausal status, use of oral contraceptives or HRT and gynecological surgical history.

Collection of data lasted from January 2008 to December 2010. Medical files of the PBC patients were retrieved and reviewed and data concerning their clinical parameters, liver biopsies, Mayo risk score at diagnosis and recent laboratory tests were annotated. FDR and controls after the interview were also clinically examined and tested for: alanine aminotransferase (ALT), aspartate aminotranferase (AST), alkaline phosphatase (ALP), γ-glutamine transferase (γ-GT), bilirubin, glucose, urea, cholesterol, triglycerides, high density lipoprotein (HDL), low density lipoprotein (LDL), fT3, fT4, TSH, anti-TPO and anti-TG, immunoglobulins IgA, IgG, IgM, rheumatoid factor (RF). Viral hepatitis B and C markers were assessed by ELISA.

Antinuclear antibodies (ANA) were tested by indirect immunofluorescence on Hep-2 substrate with 1/80 cut-off of positivity. Anti-mitochodrial antibodies (AMA) and anti-smooth muscle antibodies (SMA) were tested by an indirect immunofluorescence (IIFL) assay of Nova Lite TM (IFA) on Mouse Kidney & Stomach substrate (Inova Diagnostics, San Diego CA, Inc) and a titre of ≥1/40 was considered positive, according to the manufacture’ s instructions. Anti-M2 antibodies were assessed by qualitative and quantitative ELISA (AESKULISA, German). Negative was 1-12U/ml, grey zone 12-18U/ml and positive >18U/ml.

The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki (6th revision, 2008) and was approved by the Hospital’s ethical committee.

Statistical analysis

All data were evaluated in two sets of comparisons: patients with controls, FDR with controls. Comparisons were made by Student’s T-test for continuous variables, Fisher’s exact probability test and the χ2 test for the analysis of categorical variables. All variables found to be significant in the univariate analyses for PBC patients were entered into the multivariate analyses using a forward step-wise logistic regression model (0.05 for entry and 0.10 for removal probability). A p-value of <0.05 was considered statistically significant. Statistical analyses were performed, using the SPSS software package (version 18, SPSS Inc. Chicago, IL, USA).

Results

The diagnosis of all PBC patients was based on compatible clinical, immunological and histological parameters and all patients were on ursodeoxycholic acid 15 mg/kg since diagnosis. The mean age and the mean Mayo risk score at the time of diagnosis was 56.87 ± 11.42 years and 4.7 ± 1.6, respectively. A hundred and one patients (86 females) were AMA positive in a titre ≥1/40 on IIFL with M2 higher than 18 U/ml on ELISA, while 10/111 (8 females) were AMA negative. Fourty-three of the 111 PBC patients were ANA positive, 16 with MND pattern, 5 MND and peri-nuclear, 4 peri-nuclear, 1 peri-nuclear and anti-nucleolar, 12 speckled, 1 speckled and MND, 1 speckled and anti-nucleolar and 3 diffuse.

Fourteen patients were at stage IV at diagnosis (9 on liver biopsy, and 5 AMA positive with clinical evidence of portal hypertenstion that did not underwent liver biopsy). Of those to the end of the study, two died after diagnosis of HCC, one with HCC is still alive and 3 are alive with de-compensated cirrhosis.

Seventeen more patients were at stage III (2 AMA negative), 3 of those at stage IV alive at the end of the study, one de-compensated. Two dead (one AMA negative) of liver related causes, one within 30 days of OLT.

According to liver biopsy results at diagnosis, 79 (8 AMA negative) PBC patients were at an early stage (Ludwig I-II). During the study period three died of non liver related causes. Five are at stage IV (1 de-compensated) at the end of the study.

One female AMA positive patient refused to undergo biopsy for staging.

Mild piecemeal necrosis was present in 42.35% of the biopsies, moderate in 30.6% and no piece-meal in 27.05%.

Mean follow-up from diagnosis to interview was 91.7 ± 61 months. The mean age and the mean Mayo risk score at 2010 was 64.8 ± 11.9 and 4.99 ± 1.46, respectively.

Sociodemographic and lifestyle characteristics

The results of the sociodemographic at interview and lifestyle variables of the three study groups are seen in Table 1.

Table 1 Demographic, Anthropometric and Lifestyle variables of PBC patients, FDR and Controls
Full size table

From the 115 FDR 20% were siblings (8 brothers and 15 sisters), 76.5% were children (56 daughters and 32 sons) and 4 were mothers (3.5%) of PBC patients.

More PBC patients than controls had an elementary education (p = 0.01) and more FDR had a University degree (p < 0.0001). Among lifestyle factors, alcohol consumption and smoking were more frequent in FDR than controls (p < 0.0001 and p = 0.006, respectively), whereas mean BMI was found significantly lower in FDR (p = 0.003).

Prevalence of Autoimmune diseases in patients and FDR

PBC vs Controls

ANA positive were 38.7% vs 16.7% of controls (p = 0.002, OR 3.01 95% CI 1.5-6.02), SMA positive 18.8% vs 10.7% of controls (OR 1.93 95%CI 0.82-4.52), 75.5% of patients vs 9.9% of controls had elevated IgM (p < 0.0001, O.R. 27.8 95%CI 11.9-66.7), with a mean value of IgM = 330.3 ± 227.7 mg/dl vs controls IgM = 109.9 ± 50.7 mg/dl (p < 0.0001and IgM normal range 27-170 mg/dl).

The distribution of autoimmune diseases in PBC patients and controls are shown in Table 2. Among other autoimmune diseases diagnosed in PBC patients were: autoimmune thrombocytopenia in 2/111, multiple sclerosis in 1/111, celiac disease in 1/111, vitiligo in 1/111, autoimmune gastritis in 3/111, Guillain Barre syndrome in 1/111, BOOP in 1/111, vasculitis of autoimmune origin in 1/111, antiphospholipid syndrome in 1/111. Most patients had more than one autoimmune disease. There was a patient who had simultaneously rheumatoid arthritis, Sjögren syndrome, Raynaud syndrome, psoriasis, Hashimoto’s disease and eczema.

Table 2 Distribution of autoimmune diseases in PBC patients, controls and FDR
Full size table

FDR vs Controls

Four PBC patients were FDR of other patients resulting in a known familial history of 3.6%. Occurrence of PBC was diagnosed in another 7 FDR (6.3%) (Table 3). Five AMA positive among them underwent liver biopsy that confirmed PBC histologically (4 FDR stage I, 1 FDR stage III). The sixth was a 76 year old AMA + sister with elevated cholestatic liver enzymes that did not consent to undergo liver biopsy. An AMA negative, ANA positive 29 year old daughter with elevation of γ-GT, had a liver biopsy compatible with stage I PBC, i.e. Autoimmune Cholangitis. Addition of the 4 already known cases of familial PBC to the 7 new confirmed cases, results in a 9.9% prevalence of familial PBC in Crete. No autoantibodies were identified in the control group.

Table 3 Laboratory features of FDR with abnormal LFT
Full size table

30.8% of FDR had elevated IgM levels above 170 mg/dl, mean value 151.6 ± 136.1 mg/dl (range 23-1360 mg/dl) vs 9.9% of the controls, mean IgM value 109.9 ± 50.7, range (24.8-303 mg/dl), p = 0.009. Increased IgG levels above 1545 mg/dl, were also observed in 12.5% of FDR, mean value 1220.8 ± 263.3 compared to 1.2% of controls, mean value 1028.4 ± 266.2, p = 0.004. ANA (23.1% vs 16.7%) and SMA positive autoantibodies did not differ significantly between the two groups (16.7% vs 10.7%) p = 0.285 for both.

The distribution of other than PBC autoimmune diseases in FDR are shown in Table 2. Overall in 35 FDR (30.4%) at least one autoimmune disease (including the PBC) was identified.

Other associated diseases

PBC patients vs Controls

Dyslipidaemia (69.4% vs. 40.9%, p < 0.0001 OR 3.27, 95% CI 1.95-5.49), and thyroid gland dysfunction other than autoimmune thyroiditis (44.1% vs. 26.2%, p = 0.003, OR 2.23, 95% CI 1.32-3.76) were significantly higher in patients, compared to controls. There were significant differences in the mean cholesterol value (224.5 ± 57.7 mg/dl vs 188.1 ± 41.2 mg/dl p < 0.0001), but not in the mean values of triglycerides between patients and controls (126 ± 45 mg/dl vs. 140.1 ± 73.75 mg/dl).

Diagnosis of HT, DM, CAD, peripheral vasculopathy, asthma/COPD and allergies did not differ significantly between the two groups. UTIs did not differ between groups (p = 0.065) and more than 10 UTI/lifetime were reported by 9.7% patients vs. 6.9% of the controls (p = 0.187).

Compared to controls, patients reported more often a history of cancer (8.1% vs. 2%, p = 0.033, OR 4.29 95% CI 1.13-16.13) and the distribution was: 3 HCC, 1 breast cancer, 1 colon cancer, 1 endometrial cancer, 1 gastric MALT lymphoma, 1 basal cell skin carcinoma, 1 of unknown origin vs. 2 breast and 1 pancreatic cancer. Cholecystectomy differed significantly between the two groups (18.9% vs. 7.4%, p = 0.007 OR 2.92 95% CI 1.35-6.37) while tonsillectomy, appendectomy, and thyroid gland surgery did not.

FDR vs Controls

Abnormal liver enzymes test were found in 23/115 FDR. Their details are seen in Table 3. All relatives were negative for B and C hepatitis viral markers while drug-induced hepatotoxicity was excluded, as they had not used any medicine or herbal substance for the past six months.

Forty two FDR reported the presence of thyroid gland dysfunction that did not differ between FDR and controls (36.5% vs 26.2%, p = 0.081). HT (p < 0.0001, OR, 0.12; 95% CI: 0.058-0.23), DM (p < 0.0001, OR, 0.17; 95% CI: 0.063-0.44) and CAD (p = 0.046, OR, 0.14; 95% CI: 0.02-1.09) were found more often in controls, a fact justified by the age factor. Nonetheless, FDR despite their younger age, had hyperlipidaemia (elevation in total cholesterol and/or triglycerides, or recent treatment with statins) more frequently than controls (60% vs 40.9%, p = 0.003). Peripheral vasculopathy, asthma/COPD and allergies did not differ significantly between the two groups (p = 0.141, p = 0.195 and p = 0.42 respectively). The frequency of aeroallergens, as allergic factor was greater in FDR, while penicillin and b-lactams allergies were reported more often among controls (p = 0.031). Compared to FDR, controls reported significantly higher prevalence of UTI (p = 0.002), kidney stones and cysts (p = 0.038). A daughter and a mother of 2 PBC patients were diagnosed with ovarian and breast cancer respectively. Surgical history including tonsillectomy (14.8% vs 12.1%, p = 0.864, OR, 0.9; 95% CI: 0.46-1.78) appendectomy (11.3% vs 18.8%, p = 0.123, OR, 0.55; 95% CI: 0.27-1.12), cholecystectomy (7% vs 7.4%, p = 1, OR, 0.94; 95% CI: 0.36-2.41) and thyroid gland surgery (7% vs 3.4%, p = 0.252, OR, 2.15; 95% CI: 0.68-6.76) did not differ significantly between FDR and controls.

Reproductive history of PBC patients, FDR and controls

Results and comparisons among the three groups are shown in Table 4. All parameters with significant differences between FDR and controls can be explained by the difference between the menopausal women in the two groups (i.e. 18 menopausal FDR vs. 108 controls) as expected by the age difference.

Table 4 Reproductive history of female PBC patients, FDR and Controls
Full size table

Multivariate analysis

Variables entered in the multivariate analysis for all patients were rural residence, primary educational level, RA, Sjogren syndrome, Raynaud syndrome, Hashimoto, the presence of at least another autoimmune disease, thyroid gland dysfunction, cholecystectomy and cancer. Primary educational level (AOR 2.304, 95% CI 1.024-5.181), cholecystectomy (AOR 2.927, 95% CI 1.347-6.362) and the presence of at least another autoimmune disease (AOR 3.318, 95% CI 1.177-6.22) were a gender- independent risk factor associated with PBC.

Discussion

In this case control study we report the prevalence of autoimmune and other co morbidities as well as the sociodemographic and lifestyle factors associated with PBC in an ethnically homogeneous and geographically defined group of PBC patients. The definite advantages of population isolates, such as the Cretan population, referring to more uniform environment, genetic homogeneity and low migration rates give an added value at the study.

The limitation of the study, as with all studies that use questionnaires, is that it comprehends the risk of putative reports that lead to bias. We lowered this risk by interviewing all participants by the same doctor and validated the collected data for the PBC patients, by patient record review.

The potentially low percentage of patients that agreed to participate in the study (56.6%), could be justified by Cretan cultural ethics and prejudices, the advanced age at diagnosis, rural residence (almost half of our patients) and their low educational level. The same reasons could also explain the less frequent use of hair dye and nail polish in our patients. Indeed our findings do not support previous findings in other populations indicating nail polish or hair dye [4, 25, 27] as a putative risk for PBC.

The educational level of our PBC patients was lower than the controls and of that reported in other populations [4, 25] and was found to be an independent predictor for the disease. Previous American and French studies reported lower BMI of PBC patients compared with their controls [4, 25] but this was not found in the Cretan population. Most patients in this study although not different from controls were overweight. Current or previous smoking did not differ between PBC and controls but the passive exposure to tobacco smoke was not investigated.

Similar prevalence of autoimmune diseases in our controls with those of the American study (13.4% vs 13%) was found. However our patients had an even higher prevalence of at least one autoimmune disease compared to theirs (36.9% vs 32%) [7]. As in previous studies [7, 25] the most frequent autoimmune diseases with very high OR in univariate analysis were autoimmune thyroiditis, Raynaud and Sjögren syndrome, all found to be significantly associated with PBC while the presence of at least another autoimmune disease was found to be an independent risk factor for the disease.

FDR also had a significant higher prevalence of autoimmune diseases (30.4%). The most frequent autoimmune disease found in FDR was Hashimoto (13.9%), followed by PBC (9.9%). Indeed the known prevalence of PBC in FDR in Crete prior to the study was 3.6% similar to the 6% reported in the US study, the 5% in UK patients [7], the 4% in France and the 5.1% in Japan [28]. Nonetheless the familial screening of FDR during the study raised the PBC prevalence in FDR to 9.9%, a figure closer to the 9% reported by Lazaridis et al [5].

Although most patients were overweight and with elevated cholesterol, DM, HT, and CAD did not differ between patients and controls in accordance with the hypothesis that PBC patients may be protected against the metabolic syndrome, due to the prevention of LDL oxidation by lipoprotein X, the antioxidant effects of bilirubin and/or the elevated levels of adiponectin [29–31].

Similarly FDR as expected by the age factor had lower prevalence of HT, DM and CAD nonetheless hyperlipidemia was significantly increased compared to controls.

We did not confirm UTI prevalence in patients compared to controls, but the limitation of self-reporting, might interfere, creating misinterpretations. Tonsillectomy was not associated with PBC. By contrast cholecystectomy was, in accordance with the French findings [4].

More patients than controls reported malignancies, one third of which were HCC, giving an OR 4.29 (95%CI 1.13-16.13). This is in agreement with the increased risk for malignancies found in 212 Greek patients previously reported [32].

Non significant difference in reproductive history of PBC patients was identified.

Conclusions

In conclusion, this study has demonstrated that hyperlipidaemia and autoimmune diseases were significantly increased not only in PBC patients as expected, but also in their younger FDR compared to controls. Primary educational level, cholecystectomy and the presence of at least another autoimmune disease were found to be putative risk factors for PBC in our Greek population. The increased prevalence of malignancies previously reported was also confirmed in our study.

Given the high occurrence of familial PBC, the screening of PBC family members with AMA, especially those with at least another autoimmune disease, could be suggested in our population in order to diagnose and eventually treat the disease at an earlier stage.