We sought to identify predictors of outcome for people living with autoimmune hepatitis (AIH). We evaluated the clinical course of people with AIH across 11 Canadian centres. Biochemical changes were analysed using linear mixed-effect and logistic regression. Clinical outcome was dynamically modelled using time-varying Cox proportional hazard modelling and landmark analysis. In 691 patients (median age 49 years, 75.4% female), with a median follow-up of 6 years (25th-75th percentile, 2.5-11), 118 clinical events occurred. Alanine aminotransferase (ALT) normalisation occurred in 63.8% of the cohort by 12-months. Older age at diagnosis (odd ratio [OR] 1.19, 95% CI 1.06-1.35) and female sex (OR 1.94, 95% CI 1.18-3.19) were associated with ALT normalisation at 6 months, whilst baseline cirrhosis status was associated with reduced chance of normalisation at 12 months (OR 0.52, 95% CI 0.33-0.82). Baseline total bilirubin, aminotransferases, and immunoglobulin G (IgG) values, as well as initial prednisone dose, did not predict average ALT reduction. At baseline, older age (hazard ratio [HR] 1.25, 95% CI 1.12-1.40), cirrhosis at diagnosis (HR 3.67, 95% CI 2.48-5.43), and elevated baseline total bilirubin (HR 1.36, 95% CI 1.17-1.58) increased risk of clinical events. Prolonged elevations in ALT (HR 1.07, 95% CI 1.00-1.13) and aspartate aminotransferase (HR 1.13, 95% CI 1.06-1.21), but not IgG (HR 1.01, 95% CI 0.95-1.07), were associated with higher risk of clinical events. Higher ALT at 6 months was associated with worse clinical event-free survival. In people living with AIH, sustained elevated aminotransferase values, but not IgG, are associated with poorer long-term outcomes. Biochemical response and long-term survival are not associated with starting prednisone dose. Using clinical data from multiple Canadian liver clinics treating autoimmune hepatitis (AIH), we evaluate treatment response and clinical outcomes. For the first time, we apply mixed-effect and time-varying survival statistical methods to rigorously examine treatment response and the impact of fluctuating liver biochemistry on clinical event-free survival. Key to the study impact, our data is 'real-world', represents a diverse population across Canada, uses continuous measurements over follow-up, and our findings help inform risk stratification of patients. We provide evidence for treating clinicians, as well as those developing and evaluating new therapies, to seek evidence of good treatment response by keeping aminotransferase activity values within the reference range. Our results challenge the role of IgG as a marker of treatment response and if normalisation of IgG should remain an important part of the definition of biochemical remission. Our analysis further highlights that baseline markers of disease severity may not prognosticate early treatment response. Additionally, the initial prednisone dose may be less relevant for achieving aminotransferase normalisation. This is important for patients and treating clinicians given the relevance and importance of side-effects of treatment for patients.
Read full abstract