Key messages

• Pentoxifylline probably leads to little or no difference in mortality in alcoholic hepatitis.
•  The main guidelines have not considered an important proportion of the existing evidence, so their conclusions partially disagree with this summary.

What is the evidence.
See evidence matrix  in Epistemonikos later

We found three systematic reviews [1],[2],[3] including eight randomized controlled trials reported in 13 references [4],[5],[6],[7],[8],[9],[10],[11],[12],[13],[14],[15],[16]. 

What types of patients were included

All studies included patients with severe alcoholic hepatitis, generally based on a Maddrey score > 32.

Only three studies required histological confirmation of alcoholic hepatitis as inclusion criteria [4],[11],[16].

What types of interventions were included

Pentoxifylline dose was 400 mg t.i.d. in six studies [4],[5],[7],[12],[15],[16]. One study administered 1200 mg q.d. [11] and one study did not report dosification [10]. Duration of treatment was 10 days in one study [11], 28 days in six [5],[9],[10],[14],[15],[16] and 6 months in one [12].

In four studies standard treatment did not include corticosteroids [5],[9],[10],[11] and in three studies both groups received an identical schedule of corticosteroids [4],[14],[16. One trial evaluated both alternatives (pentoxifylline versus placebo, and pentoxifylline plus corticosteroids versus corticosteroids) in different study arms [15].

What types of outcomes
were measured

To whom this evidence does and does not apply

• The studies included severe alcoholic hepatitis patients, which is the typical situation where this treatment is considered. However, it is reasonable to extrapolate these results to all patients with alcoholic hepatitis.
• This evidence does not apply to other hepatitis etiologies (e.g. non-alcoholic fatty liver disease, viral hepatitis), specially considering there is direct evidence for pentoxifylline for those conditions. This evidence should not be applied to decompensated chronic liver disease not fulfilling criteria for alcoholic hepatitis. 

• This summary only synthesized the effects on mortality, which is the outcome critical for decision-making according to the authors of this summary. However, the results are similar for other outcomes evaluated in the different reviews (e.g. renal failure).

• There is probably no benefit in adding pentoxifylline to patients with alcoholic hepatitis, receiving corticosteroids or not. The certainty of the evidence is moderate.
• It is not clear if some benefit migh exist in specific subgroups, so future reviews might address this question. However, based on the existing data this hypothesis seems unlikely.

• It is expected some variation in the decision made by different patients and doctors; it is a probably ineffective intervention, but it is also safe and not expensive. Patients and doctors putting more value on the scant probability of benefit might be inclined to use the intervention. 

• The cost/benefit judgement does not apply since the evidence shows no benefit. 

Differences between this summary and other sources

• Key messages of our summary disagree with the existing systematic reviews. In relation with older reviews concluding there might be some benefit but the evidence is no clear [1],[2], the main difference is given by the inclusion of STOPAH trial [15], which weighs more than all other studies combined in the pooled estimated effect.
• The more recent review [3] includes STOPAH and also arrives to a conclusion of possible benefit, but with low certainty of the evidence. The main difference is it addresses the studies of pentoxifylline associated to corticosteroids (versus corticosteroids alone) as a separate comparison. From the methodological point of view this would be justified if some interaction between these drugs could be anticipated (i.e. synergistic or competing) which was not observed in STOPAH. Our summary combined all studies directly comparing pentoxifylline versus plabebo/no treatment, while Singh et al. based their conclusions on indirect comparisons. Indirect comparisons provide a more precise result (narrower margin of error), but with less certainty of the evidence (because it is indirect evidence).
• Our summary is in partial agreement with the main guidelines; the European Association for the Study of the Liver suggests its use only for cases where corticosteroids cannot be employed, while the American Association for the Study of Liver Diseases and the American College of Gastroenterology suggest considering its use in more severe patients, specially if corticosteroids will be not administered. It is important to emphasize the last version of these guidelines is 2012 and 2010 respectively, so most evidence was not available at that moment [17],[18].

• The probability of future evidence changing the conclusions about the effects of pentoxifylline in alcoholic hepatitis is low because of the certainty of the evidence.
• There are several methodological approaches to the synthesis of the existing studies, so new systematic reviews might help clarify this issue. They could also explore if there are subgroups where some benefit might be expected, or if new studies are justified.
• There are no ongoing studies for this question, at least in the International Controlled Trials Registry Platform of the World Health Organization.