PBC is an increasingly common[7] potentially life-threatening cholestatic liver disease with a poorly defined aetiology. We identified a modest association with PBC for past smoking with an aOR of 1.42 (95%CI: 0.88-2.30), which did not quite reach significance. Our effect size is close to the pooled OR of 1.67 (95%CI: 1.4-1.92) identified in a meta-analysis of five case-control studies examining the association with PBC and smoking[8]. Importantly, our associations were stronger for those who smoked for longer [e.g., aOR 2.27 (95%CI: 1.12- 4.62) for those who had been smoking for 20-35 years]. This was also the case in a study in the United Kingdom, where an OR of 3.5 (95%CI: 1.9-6.3) was found for those who smoked for 20 years or more[6]. A more robust way of assessing dose-response would be to measure the history of smoking in pack years, which, unfortunately we were not able to perform in our study. We did not identify an association with marijuana use and the development of PBC.

This is the first study that has thoroughly evaluated the intake of alcohol in PBC cases from adolescence until the age of PBC diagnosis. Surprisingly, we found that alcohol has an inverse association with the development of PBC. Our study found that cases reported significantly less alcohol intake prior to PBC onset with dose-response associations, and compared to controls, cases were more likely to report that they never drank. For example, by age 21-30 years, a third of the cases said they never drank, and this was true for only 10% of the controls aOR 3.93 (95%CI: 1.74-8.89). Our associations remained after adjustment for country of birth, education level, and smoking, and remained after a further adjustment for current alcohol intake. The big question is whether this is a true finding or whether this could have resulted from recall bias due to the stigma surrounding the intake of alcohol. Those with established liver disease may be even more likely to underreport alcohol intake prior to disease onset. However, if this is a true observation, it has important implications with respect to modifiable risk factors.

There is minimal data in the literature on the association between alcohol and PBC. There have only been two previous case-control studies where this has been evaluated and alcohol was not the primary outcome measure in either of these studies. Gershwin et al[2] found no significant difference in alcohol intake between 1032 cases and 1041 controls when they evaluated those who had consumed ≥ 12/wk standard drinks over a lifetime. However, as this is a very rudimentary measure of lifetime alcohol intake, a more detailed lifetime assessment of alcohol intake is required to evaluate this further. They also found that 28% of prevalent cases were likely to have had ≥ 12/wk standard drinks in the year prior to interview compared with 50% of controls (unadjusted P < 0.0001)[2]. However, it is difficult to make a meaningful assessment of this data, as the alcohol intake period was after disease onset, where cases would have been advised to minimise alcohol intake.

Prince et al[3] compared data from more than 2400 controls with two groups of prevalent PBC cases; cases from a geographically defined epidemiology study (epidemiological cases n = 318) and from a survey of the national patient support group (foundation cases, n = 2258). They found that PBC cases from both groups were less likely to have regularly consumed alcohol over their lifetime compared with controls [OR 0.57 (95%CI: 0.39-0.83), OR 0.73 (95%CI: 0.61–0.79) for epidemiological and foundation cases respectively]. When males were excluded from the analysis, the significant difference did not persist for the foundation cases but did persist for the epidemiological cases[3]. Thus, there was a need for a more detailed study of alcohol intake prior to PBC onset.

Alcohol is thought to protect against autoimmune diseases such as rheumatoid arthritis[9,10], SLE[11,12], autoimmune thyroid disease[13], and Type I diabetes mellitus[14]. It has been suggested that moderate alcohol intake may have a protective effect on the immune system, compared with alcohol abuse or abstinence, and this may explain the protective effect on subsequent autoimmune disease development[13,15]. The mechanism as to how moderate alcohol prevents subsequent autoimmune disease development remains unclear. Potential mechanisms to explain the beneficial effects of moderate alcohol intake on the immune system function are; loss of natural killer cell activity[16], changes in immunoglobulin levels[17] and alterations in T helper 1 (Th1) and Th2 mediated immunity[18]. Thus, there is the possibility that alcohol could also be protective for PBC.

This study has several strengths. The demographic characteristics of our case cohort were similar to other published PBC cohorts in Europe[19], and the United States[2], and the participation rate of the cases was high (91%). Therefore, it is likely that our cohort is representative of other PBC populations worldwide. Our control group was matched on age (within a five-year band), and sex and was obtained from the same geographical location as cases. A limitation of our study is the selection of controls – only 12% were recruited from the electoral roll, our intended source of recruitment for all controls, but this was not possible due to this limited response to invitation. Another potential limitation of our study is possible measurement error due to respondents summarising their frequency and quantity of alcohol intake over 10 year age intervals (except for the four-year interval of 16-20 years old). It is also possible that fatigue, a symptom in up to 80% of PBC patients[20], was significant enough prior to diagnosis to reduce or even prevent alcohol intake in PBC cases.