Glucocorticosteroids for alcoholic hepatitis
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The objectives are to:
A. Evaluate the efficacy of glucocorticosteroid intervention in patients with severe, clinically overt form of alcoholic hepatitis in terms of:
1. Mortality (primary end point).
2. Symptoms.
3. Liver biochemistry.
B. Identify possible predictive factors of a positive response.
C. Evaluate the incidence of adverse events of glucocorticosteroid intervention in these patients.
Background
The term alcoholic hepatitis was first used by Becket et al. in 1961 (Beckett 1961), but reports of clinical jaundice after excessive ethanol consumption were not unusual in the early medical literature and they probably represent instances of alcoholic hepatitis (Mendenhall 1993; Jensen 1994a). Alcoholic hepatitis is a serious form of alcoholic liver disease, and it is considered the most important precursor of alcoholic cirrhosis (Sørensen 1984). However, it is not considered a pre‐requisite for the development of cirrhosis in alcoholics (Crabb 1995). The true incidence of alcoholic hepatitis, especially of the mild forms, is unknown but it appears to differ among countries (Mendenhall 1993).
There is a wide spectrum of clinical manifestations in patients with liver morphological characteristics of alcoholic hepatitis, ranging from asymptomatic forms to overt liver failure. A higher hospital mortality occurs mainly in the initial two weeks after admission (Mendenhall 1993), and the risk of repercussions vary ‐ with or without the resumption of alcohol misuse (Galambos 1972). In recently published studies, a number of variables have been proposed to predict patients' outcome. The Maddrey formula gave the best prediction of survival, followed by serum bilirubin and creatinine alone (Mendenhall 1993). The predictive value of Mallory Bodies ‐ an important hallmark for the histological diagnosis of alcoholic hepatitis ‐ is debated (Jensen 1994a; Jensen 1994b).
Several mechanisms have been proposed for the pathogenesis of alcoholic hepatitis. These include a direct toxic effect of ethanol, enhanced lipid peroxidation, perivenular hypoxia, and malnutrition. In addition, clinical observations have also suggested a potential role for autoimmune events induced by ethanol (Gluud 1985; Zetterman 1990; Jensen 1994a; Jensen 1994b). However, the reason why severe liver injury develops in just some patients with chronic alcohol consumption is unclear.
There is no widely accepted, effective treatment for patients with alcoholic hepatitis. Abstinence from alcohol and management of associated alcohol‐related problems are the most important elements of therapy (Carithers 1989). Several therapeutic agents including propylthiouracil (Orrego 1987), anabolic‐androgenic steroids (Mendenhall 1984; Gluud 1984), and glucocorticosteroids (Carithers 1989; Ramond 1992) have been evaluated, but the effectiveness of such agents has not been conclusively proved to decrease mortality (Hallé1982; Gluud 1988; Carithers 1989).
The use of glucocorticosteroids in alcoholic hepatitis is based on evidence of inflammatory component of the hepatic injury and the possible role of immunological factors in its perpetuation (Carithers 1989; Thomson 1991). Some small trials have suggested that glucocorticosteroid intervention could improve short‐term survival in patients with the severe form of alcoholic hepatitis (Ramond 1992). The possibility to intervene with glucocorticosteroids in alcoholic hepatitis has gained some acceptance among gastroenterology specialists in Europe, 2% using this option always, and 66% sometimes (Gluud 1993). However, the results of a meta‐analysis taking prognostic variables into consideration do not support the routine use of glucocorticosteroids in these patients (Christensen 1995).
This systematic review was carried out to seek evidence of efficacy and potential adverse effects of glucocorticosteroid intervention in alcoholic hepatitis patients.
Objectives
The objectives are to:
A. Evaluate the efficacy of glucocorticosteroid intervention in patients with severe, clinically overt form of alcoholic hepatitis in terms of:
1. Mortality (primary end point).
2. Symptoms.
3. Liver biochemistry.
B. Identify possible predictive factors of a positive response.
C. Evaluate the incidence of adverse events of glucocorticosteroid intervention in these patients.
Methods
Criteria for considering studies for this review
Types of studies
All randomised clinical trials (RCTs) of glucocorticosteroid intervention in patients with severe, clinically overt alcoholic hepatitis.
The RCT may be unpublished or published as an article, an abstract, or a letter, and no language limitations will be applied.
The RCTs should use a proper method of randomisation. Thus, trials using quasi‐randomisation (i.e., controlled clinical trials) will be excluded.
RCTs may be double‐blind, single‐blind, or unblinded.
Types of participants
Patients found to have severe, clinically overt alcoholic hepatitis through clinical and/or biochemical criteria.
Alcoholic hepatitis will be considered severe if any of the following is found: the presence of spontaneous hepatic encephalopathy, Maddrey's discriminant‐function value (Maddrey's formula) higher than 32 [DF = 4,6 x prothrombin time (sec) + serum bilirubin (mg per dl)] (Maddrey 1978) and/or patients type C according to the Child‐Pugh criteria (Pugh 1973).
Types of interventions
Intervention with glucocorticosteroids in any dose or duration ‐ either enterally or parenterally ‐ compared with placebo or no active intervention.
Types of outcome measures
The following outcome measures will be assessed:
1. Mortality (primary end point).
2. Improvement of hepatic encephalopathy.
3. Improvement of biochemical variables
‐ Serum bilirubin
‐ Plasma prothrombin time
‐ Serum alkaline phosphatases
‐ Serum aspartate aminotransferase (AST) and/or serum alanine aminotransferase (ALT)
‐ Serum albumin
‐ Serum creatinine.
4. Factors associated with fatal outcome
‐ Age
‐ Sex
‐ Hepatic encephalopathy
‐ Ascites
‐ Maddrey's discriminant‐function
‐ Serum bilirubin
‐ Plasma prothrombin time
‐ Serum alkaline phosphatases
‐ Serum AST and/or ALT
‐ Serum albumin
‐ Serum creatinine.
5. Adverse events.
Search methods for identification of studies
Relevant RCTs will be identified by electronic searching the MEDLINE, EMBASE, LILACS, and the Cochrane Controlled Trials Register databases.
The Cochrane search strategy described in the Cochrane Collaboration Handbook will be used for the MEDLINE database search, using specific terms (alcoholic hepatitis, glucocorticosteroids, corticosteroids, steroids and MeSH terms).
The LILACS database search will use the search strategy described by Castro et al (Castro 1997), using the same search terms as used to search the MEDLINE database.
The bibliographic references of identified RCTs, textbooks, review articles and meta‐analyses will be checked in order to find RCTs not identified by the electronic searches.
One of the authors (CG) is involved in the handsearch of Liver, Journal of Hepatology, and Hepatology from the first issue and ongoing.
The principal authors of the identified RCTs will be approached and inquired about additional RCTs they might know of (please see Annex A).
Data collection and analysis
INCLUSION CRITERIA APPLICATION
Independently, two reviewers will apply the inclusion criteria to all potential studies. When a discrepancy occurs, the other reviewers will be asked for an opinion in order to reach consensus.
METHODOLOGICAL QUALITY ASSESSMENT OF INCLUDED RANDOMISED CLINICAL TRIALS
The methodological quality of the included RCTs will be evaluated independently by the reviewers. The following dimensions and criteria will be assessed in a standard way adopted from the Cochrane Collaboration Handbook and Schulz et al (Schulz 1995), as described by Gluud and Nikolova (Gluud 1998).
A. Generation of allocation sequence:
(3): Adequate sequence generation is reported using a computer random number generator or random number tables.
(2): Does not report on one of the adequate forms of generation of allocation concealment mentioned above in (A3), but mentions the randomisation method.
(1): Other methods of allocation that appear to be unbiased (e.g., minimisation).
B. Allocation concealment:
(3): Adequate measures taken to conceal allocations such as central randomisation; serially numbered, opaque, sealed envelopes; or other description that contains elements convincing of adequate concealment.
(2): Vaguely concealed trials, in which the authors either do not report an allocation concealment approach at all, or report an approach that does not fall into one of the categories in (B3).
(1): Inadequately concealed trials, in which the method of allocation was not concealed, such as alternation methods or use of case record numbers (quasi‐randomisation) ‐ such trials are going to be excluded in the present systematic review.
C. Inclusion of all randomised participants:
(3): Trials in which an intention‐to‐treat analysis is possible with only a few losses to follow‐up.
(2): Trials which report exclusions as listed in (C3) above, but exclusions are less than 10%.
(1): No reporting on exclusions, or exclusions greater then 10%, or wide differences in exclusions between groups.
D. Blinded assessment:
(3): Trials in which the double‐blind or double masked technique is used.
(2): Trials trying to control information bias (i.e., bias originating from withdrawals, co‐interventions, and/or assessment) by other methods (e.g., single blinded trials).
(1): Trials in which reduction of information bias is not employed.
DATA EXTRACTION
The following prespecified characteristics of all included RCTs will be extracted by two reviewers independently, conferring with the other reviewers when discrepancy occurs in order to reach consensus:
1. Methods: diagnostic procedures, randomisation procedure, concealment of allocation, whether patients were blinded to treatment, whether investigators were blinded to treatment.
2. Participants: age, sex, severity of the liver disease according Child‐Pugh criteria (Pugh 1973), Maddrey's discrimination‐function [DF = 4,6 x prothrombin time (sec) + serum bilirubin (mg per dl)] (Maddrey 1978), and presence of spontaneous hepatic encephalopathy.
3. Interventions: glucocorticosteroid dosage, duration of therapy, and way of administration as well as intervention in the control group.
4. Outcome: mortality, clinical and biochemical improvement, length of follow‐up, factors associated with fatal outcome.
5. Adverse events.
STATISTICAL METHODS
All analyses will be performed according to the intention‐to‐treat method, i.e., they include all randomised patients, and patients without the end point will be considered as failures.
The statistical package (RevMan) provided by the Cochrane Collaboration will be used. Depending on the presence or absence of RCTs variability (significant heterogeneity defined as p<0.1), a random effects model (DerSimonian 1986), or a fixed effects model (Peto (DeMets 1987)) will be used. In case significant heterogeneity is to be found, the potential causes for the heterogeneity will be explored by performing sensitivity analyses.
It is intended to perform sensitivity analyses with regard to the severity of alcoholic hepatitis at inclusion, adequacy of allocation method, as well as RCTs using double‐blind or investigator masked technique vs RCTs that are unblinded.
The existence of indications of publication bias will be explored.
The major end points mortality and hepatic encephalopathy will be examined adjusting for prognostic variables and their possible interaction with therapy (Christensen 1995).
