Alcoholic steatohepatitis (ASH) is an acute inflammatory liver disease associated with a poor outcome [¹]. In patients with a severe form of ASH, as defined by a Maddrey's score (also reported as Maddrey's discriminant function) ≥ 32, a 4-week course of corticosteroids significantly reduces the short-term mortality by approximately 25% [²].

Decompensated cirrhosis is a common clinical presentation of ASH in Western Europe [¹], but the diagnosis of ASH using only clinical and biological criteria is very challenging for the physician. Accordingly, ASH but also other situations such as infection, gastrointestinal bleeding, or drug-induced hepatitis are common precipitants of acute deterioration in patients with alcoholic cirrhosis, a condition referred to as acute-on-chronic liver failure [³]. Therefore, when a severe form of ASH is suspected, as assessed by the Maddrey's discriminant function [²] or the MELD score [⁴], a liver biopsy is strongly recommended to make a definite diagnosis and to guide steroid therapy.

Histologically, ASH is defined by the presence of steatosis (macro- and/or microvesicular), hepatocellular injury (ballooning, apoptosis), and infiltration of the liver lobule by polynuclear neutrophils[⁵]. Mallory-Denk hyaline bodies, megamitochondria, perisinusoidal fibrosis, mild iron deposits, some degree of ductular reaction (also described as cholangiolar proliferation corresponding to proliferation of hepatic progenitors[⁶]), and intraparenchymal cholestasis are also described[⁷,⁸] but not required for diagnosis. Equivalent terms used for intraparenchymal cholestasis include bile pigments accumulation, bilirubinostasis, intralobular cholestasis or cholestatic alcoholic hepatitis[⁶,⁷,⁹]. Liver biopsy is mostly performed in patients with presumed ASH to exclude other causes of liver disease and to confirm the presence of ASH. Whether the full spectrum of histological alterations observed in liver biopsies of patients with ASH has a clinical significance remains unclear[⁷]. Thus, the aim of this study was to explore the prognostic value of several histological features observed on liver biopsy performed early after hospital admission in a large number of patients with recent active alcohol intoxication, documented ASH and no associated sepsis.

Variables are given either as categorical (histological features) or continuous (age, bilirubin) values, and expressed as median and ranges. Rating for continuous variables was determined according to median value in the group. The Wilcoxon and Fisher's exact tests were used to compare variables between groups. Inter-rater agreement for qualitative variables was assessed with K statistics. A kappa value greater than or equal to 0.75 was considered to represent good agreement. Patients' survival was analyzed by the Kaplan-Meier method. The prognostic significance of the variables in the univariate analysis was determined using the log rank test. Rating for each variable was determined according to the median value in the group. To identify independent predictors of mortality at 3 months, we performed a multivariate logistic regression analysis using a Cox proportional hazard model. A 2-sided P value of < 0.05 was considered statistically significant. All statistical tests were performed by using the Statistical Package for the Social Sciences, version 10.0 for Windows (SPSS, Chicago, IL, USA).

The prominent finding of this large retrospective study is the presence of marked intraparenchymal cholestasis on liver biopsy as an independent predictor of survival, along with age and the Maddrey's score. Among other histological lesions commonly observed in ASH[²⁰] bilirubinostasis was also the sole predictor of outcome. Interpretation of intrahepatic cholestasis is especially challenging in patients with decompensated cirrhosis at risk of developing biliary tract disease, infection or sepsis. In the latter situation, intrahepatic cholestasis is a prominent finding[¹³]. Having reasonably excluded the role of bile duct lesions or concomitant sepsis with the complete work-up performed at admission, intraparenchymal cholestasis can be considered as a lesion associated with ASH, as previously suggested[²¹]. Our results are in line with the study by Nissenbaum et al.[⁹] who reported lobular cholestasis in 38% of patients that correlated to malnutrition and a poor clinical outcome. In a recent cohort of alcoholics presenting as acute-on-chronic liver failure[⁶], a majority of whom reached the histological criteria for ASH, bilirubinostasis within the ductular reaction predicted in-hospital mortality. In this study, however, infection was documented in a substantial number of patients at time or very early (< 48 h) after hospital admission, and thus a role for sepsis in the development of intrahepatic cholestasis is difficult to rule out[¹³]. In our study, having excluded patients with sepsis at admission and focused our analysis to cholestasis in the liver lobule, we believe our data truly reflect lesions associated with ASH. Nevertheless, eleven patients (4%) developed infection during follow-up and died of sepsis, a frequent complication of severe ASH treated with steroids[¹⁵]. This incidence is low compared to recent data[¹⁴], and may result from a proportion of non severe ASH in our cohort as well as the implementation of antibioprophylaxis in steroid-treated patients.

The development of jaundice and elevated serum bilirubin is common in decompensated alcoholic liver disease associated with ASH[¹]. Accordingly, serum bilirubin is integrated in several prognostic models[⁴,¹⁷,²²] that guide therapeutic decisions. In our study, most patients with high serum bilirubin, but not all, showed marked lobular cholestasis. Nevertheless, our data are in line with results from previous studies[⁶,⁹,²¹] suggesting that histology provides additional prognostic information.

The pathophysiology of cholestasis in ASH is still uncertain. Proposed mechanisms include compression of intrahepatic biliary radicals by swollen and ballooned hepatocytes[²³] or interference with basolateral and intracellular transport of bile acids[²⁴] which may in turn induce cholestasis[²⁵]. Accordingly, a particular bile acid pattern (elevated chenodeoxycholic and low deoxycholic acid serum levels) has been reported in patients with ASH [²⁶], that correlated with histological features in general, but not with cholestasis in particular. This apparent dissociation between serum bile acids, bilirubin and tissue cholestasis has been observed in almost forty percent of the cohort described by Nissenbaum et al.[⁹]. Supplementation of ursodeoxycholic acid in patients with alcoholic cirrhosis and high serum bilirubin is associated with a decrease in biological tests of cholestasis [²⁷] but histological changes are not described. Whether intracellular and canalicular trafficking of bile may be altered in ASH is a valid hypothesis. Both sepsis and ASH are characterized by a dysregulated inflammatory reaction with production of pro-inflammatory cytokines[¹⁰,¹⁶,²⁸]. In both conditions endotoxemia may be detectable, TNFα is elevated, and bile flow is impaired at the cellular level[¹³,²⁹,³⁰]. In a subgroup of our patients, however, we were unable to demonstrate any relationship between bilirubinostasis and TNFα serum levels. Whether serum levels of cytokines truly reflect their biological activity within the hepatocyte is questionable.

Except for lobular cholestasis, other histological features were not associated with clinical outcome. Steatosis results from recent, heavy alcohol intake[¹⁶,²⁰]. Major steatotic changes were present in the majority of patients, associated with recent and heavy alcohol intoxication. In alcoholic liver disease, the severity of steatosis predicts the development of cirrhosis over a 4-year period[³¹], but the short term significance of this feature in ASH is unknown. In our personal experience, steatosis is reduced and even disappears within weeks following decompensation in case of sustained alcohol abstinence. Marked hepatocellular damage as assessed by frequent Mallory-Denk bodies and ballooned hepatocytes was highly prevalent (> 70%) but not significantly related to short term mortality (p = 0.086). In a recent study on 54 alcoholic patients presenting as acute-on-chronic liver failure [³²], the majority of whom were admitted to the intensive care unit and developed infection, the presence of Mallory-Denk bodies and ductular bilirubinostasis were predictors of in-hospital mortality. In a large cohort of patients with ASH treated with steroids, Mathurin et al [¹⁷] reported a better 1-year survival in patients with marked liver neutrophilic infiltration compared to those without this histological finding. Our results didn't confirm this association, and we hypothesize that an older age (55 years in our study versus 48 years in the study by Mathurin[¹⁷]), previously suggested to play a role in glucocorticoid efficacy, as well as inclusion of 27% of patients with non severe ASH in our cohort may explain this difference. Elevated liver iron is associated with accelerated fibrosis in alcoholics [³³] and increased long term mortality in patients with alcoholic cirrhosis [³⁴]. However, over a short time period in patients with ASH and cirrhosis, marked iron deposits didn't influence survival. Hepatocellular regeneration is an important point to consider in the evolution of decompensated liver disease. Extensive necrosis and apoptosis, replicative senescence of mature hepatocytes, toxicity of alcohol and metabolites, deleterious effect of cholestasis[³⁵], may all participate in poor organ repair mechanisms in cirrhosis with superimposed ASH[³⁶]. The intensity of the ductular reaction in the periportal region, considered as representative of regeneration processes[³⁷], was not able to discriminate survivors and non survivors, consistent with recent data[⁶].

The strengths of our study is a detailed analysis of a large group of patients with a well-defined alcoholic liver disease based on histological findings obtained very early after hospital admission, presenting either in a severe or non severe form of ASH, in whom we described the outcome 3 months after clinical decompensation. Having performed a semi-quantitative analysis of the severity of lesions, we demonstrated a significant correlation between one of these features and clinical outcome. Accordingly, intrahepatic cholestasis is an independent predictor of 3-month mortality, together with age and the Maddrey's score, a commonly used robust parameter that includes serum bilirubin and prothrombin time[³⁸]. We acknowledge that our study suffers from some limitations. First, due to missing date, we were not able to include modern prognostic markers such as the Lille[²²] or the Glasgow[³⁹] scores, nor the C-reactive protein. Whether intraparenchymal cholestasis combined with the short-term evolution of serum bilirubin included in the Lille score brings additional information on prognosis remained to be explored. Secondly, our assessment of intrahepatic cholestasis may be suboptimal with regards to the semi-quantitative analysis, and the criticism may be raised that conversion of a subjective observation (i.e. lesion present in more or less than 50% of the biopsy) into a categorical variable is arbitrary and subject to inconsistency. However, we have tested the reproducibility of the scoring system and demonstrated an excellent inter observer agreement between the two histopathologists. Thirdly, the prevalence of histological lesions in alcoholic liver disease may be influenced by the timing of liver biopsy in relation to clinical deterioration, as features such as steatosis, hepatocyte ballooning and Malloy-Denk bodies tend to be less prominent over time[⁴⁰]. Therefore, it is questionable whether intralobular cholestasis would still be relevant if the liver biopsy had been performed later in the course of the disease.

This detailed analysis of histological features observed on liver biopsies of patients with ASH demonstrates that among the full spectrum of lesions observed in alcoholic liver disease, only marked intralobular cholestasis is closely associated with the short term clinical outcome. It may be hypothesized that incorporation of this variable into existing disease severity scores for ASH would improve their performance.

ASH: alcoholic steatohepatitis; MELD: model for end-stage liver disease; Maddrey's DF: Maddrey's discriminant function; INR: international normalized ratio; TNFα: tumor necrosis factor alpha

The authors declare that they have no competing interests.

LS designed and conceived the study, performed liver biopsies, was responsible for acquisition and interpretation of data, drafted the manuscript. LRB participated to the conception of the study, interpreted liver biopsy, drafted the manuscript. MG participated to the conception of the study, performed liver biopsy interpretation, helped to draft the manuscript. AH participated to the conception and search for funding of the study, helped to draft the manuscript. EG participated to the conception and coordination of the study, performed the statistical analysis, participated to draft the manuscript.

All authors read and approved the final version of the manuscript.

This study was supported by an unrestricted grant from the Foundation for Liver and Gut Studies (FLAGS) in Geneva. The authors thank Nicolas Goossens for his help in proof reading the manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-230X/11/115/prepub