Introduction We have previously1 reported presence of persisting necro-inflammation (Ishak score (NIS) >3), despite biochemical remission, in 38% of AIH patients; this was independently associated with mortality.

Aim We aimed here to assess outcome further in these patients and to identify demographic, disease and treatment-related factors associated with death or for liver transplantation (OLT).

Patients and Methods 50 patients with AIH by IAIHG criteria (16 probable, 34 definite, 39 women, median age 52 years), who had achieved biochemical remission (ALT<2× upper normal limit, median (range) 29 U/l (9–84)) for 1 (0.5–5.5) years on immunosupression (IS) treatment, but had NIS >3 on remission biopsy (performed 2.1 (1–14) years, following diagnostic biopsy. After remission biopsy, 30 patients were given routine IS (Prednisolone withdrawal plus increased Azathioprine to 2 mg/kg BW/day) and 20 received altered/enhanced IS (increased Azathioprine with continuation of Prednisolone (n=11) or change to another IS drug n=9).

Results 16 patients died, six due to liver disease (2 HCC, 3 liver failure, 1 portal vein thrombosis). 2/6 died post OLT. 10- and 20-year mortality was 11+6% and 65+11% (all cause death/OLT) and 5+4% and 32+12 (liver related death/OLT). 11 patients had cirrhosis at diagnosis; in the remainder, de-novo cirrhosis development rate was 37+12% after 10 years. In Cox regression analysis, all-cause death/OLT was associated with older age (p=0.029) but with no other parameter, including NIS or fibrosis on follow-up biopsy, whether subsequently given routine or altered/enhanced IS, or whether or not Prednisolone was continued. Likewise, no parameters were associated with liver-related death/OLT. 18 patients underwent further remission biopsy 2.3 (1–5.5) years later. NIS fell slightly, compared to first remission biopsy (median 5 to 4 p=0.011) and fell to <3 in eight patients, all of whom survived. Fibrosis stage was unchanged (median 3 to 3).

Conclusion Persistent necroinflammation in AIH is associated with substantial mortality. However (apart from age) we could not identify predictors of mortality and specifically, could not relate outcome to alterations in IS. More intensive IS strategies may be required for these patients.