Hepatic Encephalopathy and MELD-Na Predict Treatment Benefit in Autoimmune Hepatitis-related Decompensated Cirrhosis.

BACKGROUND & AIMS: Management of patients with autoimmune hepatitis (AIH)-related decompensated cirrhosis is challenging because of the risk of treatment-related complications and lack of clinical recommendations. We investigated the predictive factors for treatment benefit in AIH-related decompensated cirrhosis at diagnosis and developed an algorithm to guide treatment decisions in clinical practice.

METHODS

This retrospective, international, multicenter study included 232 patients with histologically confirmed AIH-related decompensated cirrhosis at diagnosis. The sub-hazard ratio (SHR) of mortality was determined by competing risk analysis, considering liver transplantation (LT) as competing event. A decision tree analysis was used to develop a treatment algorithm.

RESULTS

At diagnosis, 89% of patients had ascites, and 41% had overt hepatic encephalopathy (OHE). Treated patients (n = 214; 92%) had higher aminotransferases, bilirubin, and modified hepatic activity index. The SHR of mortality was lower in treated patients (0.438; 95% confidence interval [CI], 0.196-0.981; P = .045). Patients without OHE grade 3/4 and Model for End-Stage Liver Disease-Sodium (MELD-Na) ≤28 at diagnosis were more likely to benefit from treatment. In these patients, a decline in MELD-Na ≥11 after 4 weeks of treatment had a 100% negative predictive value for death/LT. Forty-nine percent of treated patients recompensated during follow-up. Twenty percent of patients had to discontinue treatment, 65% during the first 4 weeks, and only 4% due to infectious complications. OHE ≥grade 2 and MELD-Na at diagnosis predicted the need for treatment discontinuation.

CONCLUSIONS

Immunosuppression is beneficial in patients with AIH-related decompensated cirrhosis and active disease. OHE and MELD-Na at diagnosis, along with a decline in MELD-Na at 4 weeks of treatment, are the most important determinants of outcome and can guide treatment decisions.