Letter: Predictors of Corticosteroid Response in Alcohol-Related Hepatitis—Authors' Reply

Abstract

Severe alcohol-associated hepatitis (AH) is a condition that bears a high short-term mortality [1], and corticosteroids are currently the only available therapy for patients with this disease [2]. However, given the well-documented adverse effects associated with corticosteroid use, it is clinically relevant to identify patients who will benefit the most from this treatment [3]. In our recent study [4], through a multinational analysis, we evaluated the performance of different dynamic models (defined as those using laboratory data from at least two time-points) to predict 30- and 90-day mortality in patients with severe AH. Our results revealed that the Lille day 7 score (Lille-7) was the most accurate model for predicting both 30-day and 90-day mortality. The Lille day 4 score (Lille-4) and the Trajectory of serum bilirubin (TSB) also demonstrated moderate predictive value. Interestingly, no significant differences were found when comparing Lille-7, Lille-4 and TSB.

We appreciate the letter from Forrest et al. regarding our paper and their comments on the prognostic utility of the delta neutrophil-to-lymphocyte ratio (NLR) in patients with AH [5]. Although the original study by Forrest et al. [6], on the neutrophil-to-lymphocyte ratio (NLR) was primarily designed to identify patients who were likely to benefit from corticosteroid treatment, the use of delta NLR has been explored in multiple disease scenarios with significant inflammatory components, such as AH [7]. In fact, its utility has been assessed in various contexts, including living donor liver transplantation and graft survival [8], as well as its association with increased mortality in patients with hepatocellular carcinoma [9]. Furthermore, a study conducted in France that included 116 patients with cirrhosis admitted to the ICU found that the use of delta NLR was also an independent predictor of mortality at 28 days [10]. However, when specifically evaluated in AH in our study, this score did not demonstrate predictive power for mortality at 30 or 90 days, particularly in patients receiving corticosteroids.

The use of TSB as a predictor of mortality had been validated in AH, specifically in patients before corticosteroid treatment [11]. In our study, when using this model as a dynamic score during the use of steroids, it was useful to predict mortality at 30 and 90 days. Interestingly, no significant differences were observed when comparing TSB to Lille-7, and it also demonstrated comparable performance to Lille-4. Both TSB and Lille-4 are valuable tools for risk stratification in patients with severe AH. Thus, although Lille-7 remains the most validated dynamic score, shortening the assessment period to 4 days may have a good discriminatory ability, reducing unnecessary exposure to corticosteroids. Finally, there is a clear need to identify other dynamic or static models to further improve risk stratification in this challenging condition. Future research should aim at exploring and refining these models to predict patient outcomes better and optimise treatment strategies in patients with severe alcohol-associated hepatitis.