Pentoxifylline for alcoholic hepatitis

Abstract

Background: Alcoholic hepatitis is a life-threatening disease, with an average mortality of approximately 40%. There is no widely accepted, effective treatment for alcoholic hepatitis. Pentoxifylline is used to treat alcoholic hepatitis, but there has been no systematic review to assess its effects.

Objectives: To assess the benefits and harms of pentoxifylline in alcoholic hepatitis.

Search strategy: The Cochrane Hepato-Biliary Group Controlled Trials Register, The Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, Science Citation Index Expanded, LILACS, clinicaltrials.gov, and full text searches were conducted until August 2009. Manufacturers and authors were contacted.

Selection criteria: All randomised clinical trials of pentoxifylline in participants with alcoholic hepatitis compared to control were selected for inclusion.

Data collection and analysis: Two authors extracted data and evaluated the risk of bias. RevMan Analysis was used for statistical analysis of dichotomous data with risk ratio (RR) and of continuous data with mean difference (MD), both with 95% confidence intervals (CI). Trial sequential analysis (TSA) was also used for statistical analysis of dichotomous and continuous data in order to control for random error. Where data were only available from one trial, we used Fisher's exact test or Student's t-test.

Main results: Five trials, with a total of 336 randomised participants, were included. A total of 105 participants (31%) died. Of the five included trials, four (80%) had a high risk of bias. Meta-analysis using all five trials showed that pentoxifylline reduced mortality compared with control (RR 0.64; 95% CI 0.46 to 0.89). However, this result was not supported by trial sequential analysis, which adjusts for multiple testing on accumulating data. Furthermore, four of the five trials were judged to have a high risk of bias, thus risking an overestimated intervention effect. Meta-analysis showed that pentoxifylline reduced the hepatic-related mortality due to hepatorenal syndrome (RR 0.40; 95% CI 0.22 to 0.71), but trial sequential analysis did not support this result. Data from one trial suggests that pentoxifylline may increase the occurrence of serious and non-serious adverse events compared to control.

Authors' conclusions: The current available data may indicate a possible positive intervention effect of pentoxifylline on all-cause mortality and mortality due to hepatorenal syndrome, and conversely, an increase in serious and non-serious adverse events. However, the evidence is not firm; no conclusions can be drawn regarding whether pentoxifylline has a positive, negative, or neutral effect on participants with alcoholic hepatitis.

Conflict of interest statement

None known.

Figures

1

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

3

Trial sequential analysis of the cumulative meta‐analysis of the effect of pentoxifylline on all‐cause mortality in participants with alcoholic hepatitis. The required information size of 1169 is calculated based on an a priori intervention effect of 20% (APHIS), a risk of type 1 error of 5%, and a power of 80%. The event rate in the control group is 39%, which is based on a meta‐analytic estimate of the control event rate of all the included trials. Although the cumulated z‐curve (blue curve) crosses the traditional boundary of 5% significance (horizontal red line), it does not cross the trial sequential monitoring boundary (red curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data.

4

Trial sequential analysis of the cumulative meta‐analysis of the effect of pentoxifylline on hepatic‐related mortality in participants with alcoholic hepatitis. The required information size of 1636 is calculated based on an a priori intervention effect of 20% (APHIS), a risk of type 1 error of 5% and a power of 80%. The event rate in the control group is 38%, which is based on a meta‐analytic estimate of the control event rate of all the included trials. Although the cumulated z‐curve (blue curve) crosses the traditional boundary of 5% significance (horizontal red line), it does not cross the trial sequential monitoring boundary (red curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data.

5

Trial sequential analysis of the cumulative meta‐analysis of the effect of pentoxifylline on serum creatinine in participants with alcoholic hepatitis. The required information size of 252 is calculated based on an intervention effect of 0.25 (mg/dl) (APHIS), a risk of type 1 error of 5% and a power of 80%. The cumulated z‐curve (blue curve) crosses the trial sequential monitoring boundary implying that there is firm evidence for a beneficial effect of 0.25 (mg/dl) decrease in serum creatinine when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data.

6

Trial sequential analysis of the cumulative meta‐analysis of the effect of pentoxifylline on serum bilirubin in participants with alcoholic hepatitis. The trial sequential monitoring boundary is not calculated because the actual information size is less than 1% of the information size required. This is calculated based on an intervention effect of 1.00 (mg/dl) suggested by the one trial with low risk of bias.

7

Trial sequential analysis of the cumulative meta‐analysis of the effect of pentoxifylline on levels of TNF in participants with alcoholic hepatitis. The required information size of 318 is calculated based on an intervention effect of 4.00 pg/ml, suggested by the one trial with low risk of bias (LBHIS) (Akriviadis 2000), a risk of type 1 error of 5% and a power of 80%. The cumulated z‐curve (blue curve) does not cross the trial sequential monitoring boundary implying that there is no firm evidence for a potentially harmful effect of 4.00 pg/ml when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data.

1.1. Analysis

Comparison 1 All‐cause mortality, pentoxifylline versus control, Outcome 1 Mortality using the fixed effect model.

1.2. Analysis

Comparison 1 All‐cause mortality, pentoxifylline versus control, Outcome 2 Mortality using the random effects model.

1.3. Analysis

Comparison 1 All‐cause mortality, pentoxifylline versus control, Outcome 3 Mortality according to risk of bias.

2.1. Analysis

Comparison 2 Hepatic‐related mortality, Outcome 1 Hepatic‐related mortality using fixed‐effect model.

2.2. Analysis

Comparison 2 Hepatic‐related mortality, Outcome 2 Hepatic‐related mortality using the random‐effects model.

3.1. Analysis

Comparison 3 Sensitivity analysis, all‐cause mortality, Outcome 1 Mortality.

4.1. Analysis

Comparison 4 Hepatic‐related morbidity, pentoxifylline versus control, Outcome 1 Variceal bleeding.

5.1. Analysis

Comparison 5 Biochemical parameters, pentoxifylline versus control, Outcome 1 Serum creatinine.

5.2. Analysis

Comparison 5 Biochemical parameters, pentoxifylline versus control, Outcome 2 Serum bilirubin.

6.1. Analysis

Comparison 6 Post‐hoc outcome measures, TNF levels, Outcome 1 Tumour necrosis factor.