Introduction Alcoholic liver disease (ALD) frequently has a poor prognosis and in many populations, including in the UK, its impact is growing as a result of alcohol consumption patterns. Whereas abstinence from alcohol is clearly the optimum approach to treatment, patients need to survive the initial episode of liver disease in order to benefit. Although the trigger for ALD is self-evident, the process by which alcohol gives rise to liver injury remains unclear, precluding the informed targeting of interventions based on ameliorating pathogenic processes. Furthermore, we lack reliable biomarkers to identify patients at particular risk of poor outcome at the outset of a liver disease episode in order to target therapy. Several studies have identified the presence, in sub-sets of ALD patients, of antibodies reactive with adduct-modified self-proteins such as malondialdehyde (MDA)-adducted albumin (MDA-HSA) and self-antigens such as cytochrome P450 2E1 (Cyp2E1, arising, it is thought as a consequence of an “altered-self” mechanism). To date, however, the biological significance of these antibodies and the implications that they hold for prognosis and treatment are unclear.

Aim In this study we set out, in a serial cohort of 38 ALD patients (all cirrhotic and all continuing to consume alcohol), to address the biological significance of auto- and allo-antibody responses.

Method Patients were fully phenotyped with regard to their antibody, clinical, biochemical and histological status. Clinical follow-up was then undertaken for 5 years.

Results The presence of both MDA-HSA and Cyp2E1 reactive antibody was significantly associated with risk of death during follow-up (Cyp2E1 AUC for prediction of death during follow-up 0.78 (95% CI 0.64 to 0.93), p=0.01; MDA 0.73 (0.55–0.92, p=0.05). An optimal composite measure based on reactivity to both antigens was highly predictive of risk of death during follow-up (auc 0.83 (0.7–0.96, p=0.005)). Interestingly amongst baseline biochemical parameters only bilirubin was (weakly) predictive of death during follow-up (auc 0.74 (0.62–0.92), all other biochemical parameters p=ns). Individual histological parameters were similarly not predictive of death during follow-up.

Conclusion Antibody reactivity with allo- and auto-antigens in ALD is a predictor of poor outcome and the optimal composite risk measure warrants prospective validation in outcome series. It is unclear at present whether the association with antibody reactivity results from a pathogenetic process (immune-mediated damage driving liver injury) or occurs as a consequence of enhanced injury (increased liver damage enhancing reactivity to these antigens). Further work in this area is warranted.