TABLE 3.

Predictive prognostic features of autoimmune hepatitis (AIH)

Prognostic features	Implications	Challenges
Model scores	MELD score ≥12 points predicts treatment failure (sensitivity, 97%; specificity, 68%) (70) Unimproved UKELD predicts poor outcome (sensitivity, 85%; specificity, 68%) (71)	Not disease-specific (70,71) Low specificities for treatment failure (70,71)
Clinical phenotype	Young adults have HLA DRB1*03 more frequently than elderly patients (58% versus 23%) (72) Young adults often fail treatment (33%) (72) Elderly patients have cirrhosis (33%), HLA DRB1*04 (47%) and good response to therapy (72)	Lacks specificity (72) ‘Blunt’ prognostic tool (72) Routine HLA determinations discouraged (7,8,49)
Serological markers	Anti-SLA and relapse, 53% to 100% (76,77)	Anti-SLA infrequent in AIH (79)
	Anti-SLA and HLA DRB1*03, 83% (76,78)	Anti-SLA vary by genotype (76,79)
	Anti-SLA sensitivity for AIH, 7% to 19% (79)	Absent anti-SLA not predictive (76)
	Anti-SLA specificity for AIH, >90% (77,79) Anti-actin and α-actinin occur with clinical and histological activity, 91% (81)	Need indices based on pathogenic pathways (cytokine levels, immune cell populations) (9)
Rapidity of treatment response	Failure to improve within 2 weeks indicates high mortality (37,71) Improvement within 12 months has less cirrhosis (18%) and need for liver transplantation (2%) (51) Elderly respond more quickly than young (51)	Requires time investment to assess response (37,51,71) No pretreatment predictors (8) Need dynamic indexes at each stage of disease (8,37)

Numbers in parentheses refer to references. HLA Human leukocyte antigen; MELD Model of End-stage Liver Disease; SLA Soluble liver antigen; UKELD United Kingdom Model for End-stage Liver Disease