A Lille model for predicting the response of severe alcoholic hepatitis to corticosteroid treatment in Japanese patients

Aim Corticosteroids are the most widely used agents for the treatment of severe alcoholic hepatitis (SAH). The therapeutic effectiveness of corticosteroids is assessed by the Lille model, which has been validated well in patient cohorts in North America and Europe; however, its usefulness has not yet been confirmed independently in Japanese patients. The present study aimed to determine whether the Lille model could predict the prognosis of SAH in Japanese patients. Methods This was a retrospective cohort study including 32 SAH patients who were admitted to our institute from April 2011 to February 2018. According to the previously validated Lille model cut-off value, patients who received corticosteroids were classified as corticosteroid responders or non-responders (with responders obtaining a Lille score >= 0.45), followed by assessment for the 6-month prognosis. Results Out of 32 patients, 26 were treated with corticosteroids. The 28-day and 6-month mortality rates in the corticosteroid-treated group were 23.1% and 46.2%, respectively. The median Lille score was significantly higher in patients who died or underwent liver transplantation (0.647) than in those who survived without undergoing transplantation (0.226; P < 0.0182). The 6-month transplant-free survival rate of non-responders (Lille score >= 0.45) was significantly lower (27.3%; 95% confidence interval, 9.0-58.6%) than that of responders (Lille score < 0.45, 73.3%; 95% confidence interval, 46.7-90.0%; P = 0.0149, log-rank test). Conclusions The Lille score could be useful for predicting the 6-month prognosis of Japanese SAH patients after corticosteroid therapy.