First studied in the treatment of alcoholic cirrhosis in 1960[44], the use of glucocorticoids remains perhaps the most studied and debated intervention. Reported successes of glucocorticoids are variable and appear largely dependent on the nature of the trial.

The rationale for the use of glucocorticoids is centered upon blocking the cytotoxic and inflammatory pathways in alcoholic hepatitis. Glucocorticoids decrease circulating inflammatory cytokines such as TNF-α, ICAM-1 expression, and have demonstrated short term histologic improvement in the treatment of alcoholic hepatitis[45].

It is difficult to provide a simple summary of results for glucocorticoid trials in alcoholic hepatitis. While all trials appeared to have been controlled, few have high quality scores given the variable definition of randomization and blinding in each trial[46,47].

The trials vary by inclusion/exclusion criteria, glucocorticoid type, scoring system, length of treatment and co-interventions. The study by Mendenhall et al is particularly difficult to interpret as essentially three different intervention arms, oxandrolone, prednisolone and prednisone, are employed[48]. Lesesne et al compares prednisolone to a 1600 caloric intake diet which is below the estimated caloric needs of most hospitalized patients; his trial is not placebo controlled[49]. Furthermore, the variation in type of glucorticoid, dose, and treatment length makes it difficult to provide treatment guidelines for physicians.

We did a computerized search using the MEDLINE database from 1971 to August 2005 using the search headings of “steroids”, “corticosteroids”, “alcoholic hepatitis”, “hepatitis, alcohol”, “randomized” and “English”. We obtained additional trials by manually searching through retrieved trials and review articles. Randomized trials including corticosteroids in the treatment of alcoholic hepatitis with the outcome measure of mortality are summarized in Table 3. The trial results are summarized by percentage death in each group including number of patients with hepatic encephalopathy and their percentage mortality. Relative risk reduction (RRR), number needed to treat (NNT) with their associated 95% confidence intervals are calculated for trials that reported a significant benefit in survival. Given that hepatic encephalopathy is a known predictor of mortality, we have also calculated RRR and NNT for those patients; again only in trials that reported a significant survival benefit. We employed the Jadad score as an assessment of trial quality. The Jadad score is one of the few validated measures of randomized trial quality[46,50,51]. Out of a maximum score of five, points are assigned based on the method of randomization, double blinding and description of withdrawals/drop-outs.

The results are variable. Five trials reported a significant survival benefit with glucocorticoids[13-15,28,49]. RRR ranged from 67% to 86%, with NNT varying between 2 and 6. As reported in the above trials and noted in our table, glucocorticoid treatment significantly reduces mortality in patients with hepatic encephalopathy. The latter trials by Carithers et al and Ramond et al, selectively treated patients with discriminant function scores greater than 32, supporting a more discriminate use of glucocorticoids[14,15].

Of the trials which reported a non-significant benefit, there are four trials which report higher mortality in the glucocorticoid group[30,48,52,53]. While glucocorticoids are relatively benign in the short term for most patients, the remaining three trials remind clinicians that there are significant complications with their use. Blitzer et al reported a higher number of fungal infections in the steroid group contributing to the greater percentage of deaths in the steroid group when compared to placebo. However, his steroid treatment group contains a higher proportion of patients with elevated total bilirubin, when compared to placebo, which may contribute to his study result[52]. Included in our discussion of nutrition in the treatment of alcoholic hepatitis, Cabre et al’s study found 31% (11/35) mortality in the steroid group with 91% of deaths attributable to infection[54]. It is important to recognize the potentially serious infectious complications secondary to steroid treatment.

Subsequent meta-analyses, while still yielding conflicting results, began to delineate which patients would most benefit from glucorticoid treatment. The first meta-analysis on this topic, conducted by Imperiale and McCullough in 1990 (which antedates more recent trials), finds a protective efficacy of glucocorticoids in higher quality trials, particularly those that exclude patients with gastrointestinal bleeding but include patients with hepatic encephalopathy[55]. They found a protective efficacy of 34% overall (95% CI, 15%-48%) for patients with hepatic encephalopathy. Imperiale and McCullough’s study applies quality scores which are important in the setting of such trial heterogeneity. In their paper, quality scores are assigned by independent assessors. Quality scores we use in this paper are derived from Jadad et al, and have been used in assessing the quality of randomized clinical trials[46]. We do not find a significant association between high Jadad score and trial survival benefit. However, Jadad et al score does not consider baseline equivalence of compared groups, use of co-therapies and adequate potency of principal therapy. The variance of survival amongst the trials may have more to do with patient inclusion/exclusion criteria and the self-derived scoring systems than trial quality and adherence to randomization and double blinding.

Meta-analysis conducted by Christensen et al, did not find an overall treatment benefit, after attempting to control for confounding variables[56]. Controlling for confounding variables without direct access to individual study data can be difficult given the heterogeneity of prior trials. A subsequent study, by Mathurin, Mendenhall, Carithers et al[57] pooled raw data from their respective trials based on DF score (greater than 200 patients with DF ≥ 32 in placebo versus steroids) and found a survival benefit. If the DF < 32, there was a > 90% survival without steroids. The conclusions from the above study provide a more definitive treatment guideline for clinicians. In patients with DF ≥ 32, treatment with glucocorticoids improves short term, 28 d, survival with mortality decreasing from 35% in controls to 15% with steroids.

The longer term benefit of glucocorticoids are difficult to assess given the variable long- term clinical trial definitions (1.5 mo, 6 mo, 1 year)[54,48,58] and each of the existing three trials reported different outcomes: harm, no benefit and benefit[54,48,58]. It is also difficult to assess long-term benefit as alcoholic hepatitis is likely to recur unless the patient abstains.

A review of the literature supports a more discriminate use of glucocorticoids in patients with a Maddrey discriminant function score ≥ 32. If there is no evidence of gastrointestinal bleeding or infection, the frequent concomitant presence of hepatic encephalopathy provides an even stronger support for the use of glucocorticoids. A study by Mathurin et al suggests a simple method to identify patients who are most likely to respond to glucocorticoids. Patients with an ‘early change in bilirubin levels’ (ECBL), i.e. a bilirubin level at 7 d lower than the bilirubin level on the first day of treatment, were significantly more likely to survive and respond to steroid treatment[59]. Discontinuation of glucocorticoid treatment in the non-responder group, i.e. patients that did not have an ECBL at 7 d, did not appear to result in adverse events[5]. Devising methods to target the patient groups most likely to benefit are important in maximizing treatment benefit, avoiding unnecessary complications of treatment and streamlining the treatment decision process.

Glucocorticoids, while providing a benefit in a select group of patients, are not without risks and should be used with caution in patients with infectious complications and gastrointestinal bleeding. Further trials that are larger in sample size, involving multiple centers and with an active comparator, i.e. pentoxifylline, are needed to better delineate the true effect of glucocorticoids. Finally, as scoring systems are changing, repeat studies may be needed to reassess the treatment effect of glucocorticoids employing MELD and GAHS.

There is a measurable and clinically apparent decline in gonadal function in patients with alcoholic liver disease[60-62]. In 1938, administration of androgens appeared to enlarge the liver of cirrhotic rats, thereby suggesting that perhaps androgens could reverse the process of fibrosis[63]. This lead to the first clinical trial where 12 patients with alcoholic cirrhosis were injected with large doses of testosterone with 'some improvement'[64].

A 2003 Cochrane systematic review could not demonstrate a significant effect of anabolic-androgenic steroids on the mortality of patients with alcoholic liver disease[65]. Three trials[48,66,67] included in the analysis and two trials[68,69] excluded from the analysis are trials in which all participants have alcoholic hepatitis.

Reflective of the conclusions derived from the systematic review, Bonkovsky et al’s study[67] and Mendenhall et al’s 1984 study[48] did not find a significant survival advantage in the anabolic steroid group in the placebo. In Mendenhall et al’s study, patients in the oxandrolone group were treated for 30 d. They report, however, in subgroup analysis that patients with moderate hepatitis treated with oxandrolone seemed to have survival advantage 6 mo post treatment. As the subgroup analysis did not include patients that had died within the first two months of treatment, the results should be taken with some caution.

Currently, anabolic steroids are not recommended for the treatment of alcoholic hepatitis.

Pentoxifylline is a suppressor of tumor necrosis factor alpha (TNF-α), prevents leukocyte adherence to vascular endothelium and down regulates the expression of intercellular adhesion molecule-1 in monocytes[70]. The main signaling pathway is through type 1 tumor necrosis factor receptor, TNFR1. Elevated levels of TNF-α are predictive of poor survival in alcoholic hepatitis[71,72]. Other effects of this drug may contribute to its action such as its effects on membrane fluidity which determine its use in peripheral vascular disease.

First studied by McHutchison et al in 1991, in patients with severe alcoholic hepatitis (defined as DF score ≥ 32), pentoxifylline reduced the development of hepatorenal syndrome, and as a consequence mortality, in comparison to patients who received placebo[73]. A subsequent double blind placebo controlled trial, Akriviadis et al, from the same center, supports McHutchison’s findings[74]. There did not appear to be any complications as a consequence of pentoxifylline treatment. As noted by Dr. Mathurin[5], the latter study showed no improvement in liver function tests. The reported improved survival was accounted for by a reduction in the development of hepatorenal syndrome in the treatment group. This finding is in sharp contrast to the glucocorticoid trials which demonstrate an improvement in liver function and in survival compared to placebo.

A small sample size, retrospective, observational study by McAvoy et al, published as an abstract, finds a treatment benefit with pentoxifylline only in patients stratified to GAHS ≥ 9, but not in patients with DF ≥ 32[75]. As raw numbers are not available at press time, it is difficult to draw a meaningful conclusion from his study.

Recommendations: Pentoxifylline may reduce mortality from hepatorenal syndrome in the setting of severe alcoholic hepatitis but further studies are needed to confirm these findings. Aside from the need for head to head comparative trials with steroids, one wonders if the combination of the two treatments might exhibit an additive benefit.

Infliximab, used in the treatment of Crohn’s disease, rheumatoid arthritis and psoriasis, is a chimeric mouse/human antibody which binds to tumor necrosis alpha, blocking its effects[76]. Preliminary trial data was encouraging. Three trials reported either better survival than predicted, improved Maddrey score or laboratory parameters[76-78]. The largest and most comprehensive trial studying the efficacy of prednisone and infliximab in the treatment of alcoholic hepatitis was terminated early when a significantly higher number of deaths occurred in the treatment group[79]. The study received some criticism for its use of high dose of infliximab and infusion protocol which varied from previous studies. In this study, investigators cited prior studies in Crohn’s[80] and rheumatoid arthritis[81] in which there did not appear to be a relationship between dose of infliximab and rate of infection[78]. Furthermore, Dr. Naveau contends, perhaps infliximab is not the TNF-α blocking agent for alcoholic hepatitis (Table 4).

An open label uncontrolled pilot study on etanercept in the treatment of moderate to severe alcoholic hepatitis was completed[82]. Of the 13 patients treated, 7 had Maddrey DF greater than 32 and two of the seven died within 32 d. Etanercept was discontinued in 3 patients secondary to infection, hepatorenal decompensation and gastrointestinal bleeding. Therefore, there is no particular evidence one way or the other to suggest a beneficial or detrimental effect of treatment. As etanercept targets soluble TNF, whereas infliximab targets both soluble and membrane bound TNF, it is uncertain what the advantages/disadvantages are of this distinction in the setting of this specific disease target.

Infliximab is not currently recommended for the treatment of alcoholic hepatitis, outside of clinical trials. Although concerns have been raised about increased risk of infection, the more disturbing aspect has been recent warnings of acute liver failure in patients with Crohn’s disease and rheumatoid arthritis treated with infliximab[83]. This risk may preclude its use in patients with underlying severe liver injury who are less capable of withstanding an additional insult to the liver.

There are multiple etiologies for weight loss and malnutrition in patients with years of alcohol abuse. Weight loss can be reflective of years of substitution of alcohol for more than 50% of other calories[84], malabsorption of dietary fat and nutrients[85] and the induction of a catabolic state resulting in skeletal muscle depletion[86].

Recently reviewed[4] parenteral and enteral nutrition, while improving liver function in a few studies in alcoholic hepatitis[87-89] has yet to demonstrate a change in clinical outcome.

Mendenhall et al have done the most extensive assessment on the effect of protein calorie malnutrition (PCM) and protein energy malnutrition (PEM) on survival and liver function. The results of their interventional studies are shown in Table 5. The observed associations between degree of malnutrition, as calculated by PCM or PEM score, and severity of liver disease[90-92] as well as improvement of survival with improved PCM score[93] serve as the basis for determining the effect of nutritional intervention on survival and liver function in alcoholic hepatitis[69,94].

Nutritional interventions such as caloric amount, type, mode and duration of supplementation vary among the trials. For example, the 1600 caloric nutritional intervention in Lesesne et al’s study is below that of most hospitalized patients. A positive correlation between nutritional intake and survival, if present, would not be expected. Much as in the case of the glucocorticoid literature, it would be difficult to provide clinical recommendations when the treatment interventions and outcomes vary. Furthermore, it is difficult to draw meaningful clinical conclusions. While nitrogen balance improves in the nutrition intervention arm, survival remains unchanged.

The majority of trials did not find a survival advantage in nutritional support. There are two trials which showed a survival advantage. Nasrallah et al’s study is smaller and both groups receive a 3000 kCal diet with protein, which is an intervention treatment in some studies. Mendenhall et al, found a later survival advantage, 6 mo post treatment, in the moderately malnourished group. This is in contrast to Mezey et al’s study which did not find a survival advantage up to two years after treatment.

Recommendations: It is important to assess nutritional status of patients in order to recognize and treat the distinct nutritional deficiencies inherent in alcoholic cirrhosis and hepatitis. At this time, however, nutritional supplementation during acute presentation of alcoholic hepatitis does not appear to affect survival.

The final histologic stage in alcoholic liver disease is cirrhosis. Found to inhibit liver fibrosis in rats[95], colchicine’s anti-fibrotic activity presented a theoretical possibility of preventing liver fibrosis in humans.

Three clinical trials in the setting of alcoholic hepatitis[96,97] and a Cochrane database review in the setting of alcoholic and non-alcoholic liver fibrosis[98] fail to find a benefit in the treatment of alcoholic hepatitis with colchicine. Recently published and not included in the cochrane review, the largest trial studying long-term colchicine in the setting of alcoholic cirrhosis did not find a therapeutic benefit when compared to placebo, in concordance with prior literature[99]. Colchicine is not currently recommended for the treatment of alcoholic hepatitis.

SAMe, produced from methionine by adenosylmethionine synthetase, is important in the metabolism of nucleic acids, structure and function of cell membranes and as a precursor of glutathione. Glutathione may be protective in alcohol induced liver injury[100]. However, in liver disease there is an impairment of enzyme activation of methionine which cannot be corrected by methionine supplementation[101]. In the setting of alcoholic hepatitis, there is a measurable decrease in hepatic methionine, SAMe and glutathione levels[102]. In animal studies, administration of SAMe increased glutathione levels, attenuated ethanol induced liver injury as well as liver injury caused by other hepatotoxins[103-106].

In a 2001 Cochrane systematic review[107], SAMe has yet to consistently demonstrate a significant beneficial effect on the mortality in the setting of alcoholic liver disease. None of the analyzed trials in the systematic review targeted patients with alcoholic hepatitis. The largest multi-center and highest Jadad quality scoring trial, by Mato et al, treated patients with alcoholic cirrhosis with SAMe for up to two years[108]. There was an overall decline in mortality in the treatment group compared to placebo, but did not reach significance. Excluding patients with Child’s C cirrhosis, however, did yield a significant mortality benefit.

There are currently two NIH funded trials studying the effect of SAMe on the mortality in the setting of alcoholic cirrhosis. There has yet to be a trial studying the effect of SAMe administration on survival in the setting of acute alcoholic hepatitis.

SAMe is currently not recommended in the treatment of acute alcoholic hepatitis.

Found to reduce hypoxic hepatocellular injury in ethanol fed rats[109], subsequent animal studies confirm PTU’s protective role against oxidative and ischemic liver injury[110]; similar hepatic injuries are found in patients with alcoholic hepatitis[111].

In a 2001 Cochrane systematic review[112], PTU did not provide a significant survival benefit in the setting of alcoholic liver disease. All of the analyzed six studies (3 of which were published only in abstract) included patients with alcoholic hepatitis[110,113-117].

Contrary to animal studies, hepatic histologic improvement with PTU administration is not replicated in clinical trial literature. PTU also does not appear to have a measurable effect on splanchnic hemodynamics in the setting of alcoholic cirrhosis[118].

While the systematic review did not find a significant association between PTU and adverse events, one trial was discontinued when higher mortality rates were observed in the PTU group[110]. Furthermore, there are case reports and several reviews on fulminant hepatic failure and hepatitis[119-123] secondary to PTU in addition to leukopenia[124] Propylthiouracil is not recommended for the treatment of alcoholic hepatitis.