Anti-tumor necrosis factor-alpha therapy in severe alcoholic hepatitis: are large randomized trials still possible?
Article Outline
- 1. Introduction
- 2. Advantage of anti-TNF and prednisone combination vs. prednisone alone in reducing Maddrey's score
- 3. Advantage of anti-TNF in reducing IL-8
- 4. Severe adverse events due to anti-TNF
- 5. The paradox of trials in AH: efficacy endpoint or adverse event?
- 6. Lessons from the old corticosteroids trials
- 7. For infliximab the question is now on infection: is there an attributable risk of ‘acute’ bacterial infection, candidiasis and/or tuberculosis?
- 8. Conclusion
- References
- Copyright
1. Introduction
Severe alcoholic hepatitis (AH) is associated with high mortality [1]. Tumor necrosis factor-alpha (TNFα) has been demonstrated to play an important role in its pathophysiology [2]. Two very interesting preliminary trials recently published in the Journal suggest that a single infusion of the anti-TNF monoclonal antibody (infliximab) is effective, which reinforces the rational for treatment [3], [4]. The authors of both articles concluded that a randomized placebo-controlled clinical trial of anti-TNF antibody in severe AH is warranted to look for a benefit in survival. At the same moment an ongoing multicenter randomized trial of infliximab in patients with severe alcoholic hepatitis has been interrupted by the French drug agency (AFSSAPS) due to a suspicion of severe adverse events, mainly infections (Sylvie Naveau, personal communication).
Knowing these facts what can we say to a clinician willing to coordinate or include patients in a new large randomized trial in order to estimate the efficacy of infliximab? Yes or No?
2. Advantage of anti-TNF and prednisone combination vs. prednisone alone in reducing Maddrey's score
Spahr et al. have randomized 20 patients with biopsy-proven severe AH (Maddrey's score >32) [3]. They received prednisone 40 mg/day for 28 days and either infliximab 5 mg/kg body weight IV or placebo at day 0. All patients received standard nutrition and vitamin supplementation during hospitalization in the study period. Patients with low-protein (<10 g/l) ascitic fluid received prophylactic norfloxacin. Histology, plasma interleukin-6 (IL-6) and interleukin-8 (IL-8) were measured at baseline and at day 10. Infliximab was well tolerated. Histology showed no significant changes. At day 28, Maddrey's score significantly improved in the infliximab–prednisone group (39 to 12, P<0.05 vs. baseline) but not in the prednisone group (44 to 22, P=NS). At day 10, IL-6 and IL-8 decreased in the infliximab–prednisone group (25 to 4.5 pg/ml, 301 to 146 pg/ml, P<0.01, P<0.05 vs. baseline, respectively). In the prednisone group, changes were not significant (38 to 16 pg/ml, 315 to 110 pg/ml). They concluded that in severe AH, infliximab–prednisone was well tolerated and associated with a significant improvement in Maddrey's score at day 28. They stated that although the size of this study does not allow comparison between groups, these promising results should encourage larger trials assessing the effects of this therapy on survival.
As stated by the authors the results were not clearly significant between groups as all the endpoints were also improved in the prednisone group. The design of a randomized trial should include a large number of patients.
3. Advantage of anti-TNF in reducing IL-8
Tilg et al. included 12 similar patients with biopsy-confirmed AH and a Maddrey discriminant factor of >32 [4]. The patients received 5 mg/kg body weight infliximab as a single 2 h infusion. Prior or concomitant steroid therapy was not allowed in this study. All patients were hospitalized and received standard treatment including treatment of alcohol withdrawal with benzodiazepines, administration of fluid, calories, vitamins and minerals and management of ascites. Standard diet contained at least 25 Kcal/kg per day administered either orally or through a nasogastric tube. Patients were observed for potential concomitant infection and medical management was individualized according to each patient's condition and complication. Patients with low-protein (<10 g/l) ascites received prophylactic norfloxacin. Serial measurements were made for various cytokines using specific ELISA assays. In four patients liver biopsy samples were available pre-treatment and on day +28 of therapy. Ten of the 12 patients are alive at a median of 15 (12–20) months. Two patients died within 30 days from septicemia. Serum bilirubin levels, Maddrey score, neutrophils count and C-reactive protein fell significantly within the first month. There was an early, though not significant, decrease in plasma levels of proinflammatory cytokines (IL-1β, IL-6, IL-8, interferon-gamma), whereas plasma levels of TNFα remained near the sensitivity limit of the assay throughout the treatment course. While TNFα mRNA expression in the liver did not change, expression of IL-8, a cytokine regulated mainly by TNFα, was almost absent on day +28. They concluded also that a randomized placebo-controlled clinical trial of anti-TNF antibody in severe AH is warranted.
As stated by the authors, there was a significant modification in biochemical endpoints between baseline and follow-up but there was no control group. The reduction of IL-8 transcription was not associated with a reduction in TNFα.
4. Severe adverse events due to anti-TNF
In October 2002 a multicenter randomized trial of infliximab in severe AH was stopped by the French drug agency (AFSSAPS). The aim of this study was to determine the superiority of three infusions of infliximab (10 mg/kg), at baseline, 2 weeks and 4 weeks associated with 4 weeks of prednisolone (40 mg per day) compared with three infusions of placebo associated with the same prednisolone regimen. The main endpoint was the 2-month mortality rate. An interim analysis was performed after the inclusion of 36 patients. There was a two-fold increase of deaths in the infliximab group vs. the placebo group. In both groups the main cause of death was infection (Sylvie Naveau, personal communication). The detailed analysis is still ongoing.
5. The paradox of trials in AH: efficacy endpoint or adverse event?
One paradox of trials performed in AH, including the Tilg and Spahr trials, is to exclude patients with a higher risk of lethal complications. If one wants to see effectiveness on the mortality, the drug must decrease the main causes of deaths, which are either infections or hemorrhage. If corticosteroids or infliximab are contra-indicated in case of bleeding or infection it seems schizophrenic to look for a reduction in the number of deaths. In the Spahr trial, patients were excluded if Maddrey's score was >55 “mainly because of the potential risk of infections”, the presence of severe renal failure (serum creatinine >170 μmol/l), the presence of uncontrolled infection or recent (<15 days) gastrointestinal hemorrhage. In the Tilg trial, patients were excluded in case of concurrent infection, gastrointestinal hemorrhage, and renal failure (creatinine >200 μmol/l).
6. Lessons from the old corticosteroids trials
Bleeding and infection were considered wrongly as contra-indications of corticosteroid therapy in patients with AH. For corticosteroids it took around 30 years to show that peptic ulcer is not a contra-indication. Curiously infection is still considered a contra-indication despite the fact that prednisone decreases mortality by reducing infection. In our experience exclusion of patients with infection and hemorrhage reduces the number of patients by 45%.
In the seventies, physicians were afraid that prednisone induced peptic ulcer in patients with cirrhosis. Harold Conn raised the controversy when he looked to the evidence justifying the exclusion of patients with a history of gastro-duodenal peptic ulcer from the trials of prednisone in AH [5], [6], [7]. Indeed 30 years later there is no reason to exclude these patients. Furthermore, the number of cases of gastrointestinal bleeding is reduced in patients treated with prednisone in comparison with placebo [6], [7], [8].
There are also no facts or rational to exclude prednisone treatment in patients with AH and infection. A randomized trials overview has not demonstrated more frequent bacterial infection in patients treated with corticosteroid vs. placebo [7]. In patients with AH receiving corticosteroids there is less death related to infection than in the placebo groups [1]. However, in the long term an increase of fungal or tuberculosis infection cannot be excluded. There is also a possibility of a rebound risk of bacterial infection after stopping corticosteroids.
The goal of the treatment in AH is to reduce mortality. In this disease, the mortality is mainly due to infection and hemorrhage, very often associated with renal failure. The largest review of mortality in the placebo group was the meta-analysis we performed with Philippe Mathurin and other colleagues [1] including individual data of four randomized trials. Altogether a total of 113 patients receiving corticosteroids and 102 patients receiving placebo had a discriminat function ≥32. At 28 days, corticosteroid-treated patients had a very significantly higher survival than placebo patients: 84.6±3.4% vs. 65.1±4.8% (P=0.001).
When we looked at the cause of death in the first year in a randomized trial of prednisolone vs. placebo, there were less deaths related to infection in the prednisolone (four out of 32) than in the placebo group (six out of 29), as well as less hemorrhage (two out of 32 vs. six out of 29, respectively) [8].
We recently looked again to the cause of death in 89 consecutive patients with the same criteria for severe AH and treated for 4 weeks with prednisone (A. Chryssostalis, personal communication). The survival was 80% at 4 weeks and 70% at 8 weeks. A total of 38 patients died. Infection was considered as the main cause of death in 13 patients (15%), hepato-renal syndrome in 12 patients (14%), and gastrointestinal hemorrhage in nine patients (10%). At admission 40 patients (45%) already had a severe complication: 25 patients (28%) had infection and 15 (17%) had gastrointestinal hemorrhage.
Another reason to include severe patients is the study with pentoxifylline, an inhibitor of TNF transcription, in AH [9]. In this large placebo-controlled study, pentoxifylline improved survival by 40% and furthermore reduced the incidence of hepato-renal syndrome by 65–70%.
7. For infliximab the question is now on infection: is there an attributable risk of ‘acute’ bacterial infection, candidiasis and/or tuberculosis?
Eventually the benefit/risk ratio of infliximab will depend mainly on the attributable risk of severe infections in patients with AH. In Crohn's disease an increased risk of tuberculosis has been observed in patients treated with infliximab [10]. The difficulty for severe liver disease is that the baseline risks of bacterial, fungal or tuberculosis infection are very elevated. Therefore, the number of patients needed to exclude a significant attributable risk of infliximab will be very large. So far, in the two trials discussed [3], [4] and the French trial a total of less than 40 cases have been followed up after one infliximab injection.
In the Spahr trial [3], two patients in the infliximab group died due to liver failure and massive hemorrhage after 51 and 28 days of treatment, respectively, and one patient in the prednisone group died due to liver failure after 91 days. Two cases of infections were observed both in the infliximab group (hospital-acquired pneumonia at day 24 of treatment, and Staphylococcus Aureus skin abscesses at day 40) and in the prednisone group (Staphylococcus Epidermidis epidural abscess at day 7, and spontaneous bacterial peritonitis at day 40). In the Tilg trial, two of the 12 patients (17%) died within the first 4 weeks of receiving infliximab. Both deaths were due to septicemia (one Candida albicans, one Staphylococcus aureus). One of those two patients developed renal failure within 2 weeks after infliximab administration and died on day +21 due to Candida septicemia. Despite initial improvement in liver function, the other patient developed staphylococcal septicemia and died subsequently from multiorgan failure after 2 weeks. For the Naveau trial, we have to wait for the detailed publication. However, the same issue of attributable severe adverse events is raised.
The second important question is the best dose of infliximab. The dose used in the Naveau trial was six times greater than in the two other studies.
8. Conclusion
From these cases we can conclude with the authors that in view of the rational and the pilot study results, a large randomized trial of infliximab is warranted. An overview of previous treatments suggests that prednisone or prednisolone must be given as the standard of care in the control group. We suggest that infection and hemorrhage should not be causes for exclusion. However, because of the risk of severe adverse events (already described in Crohn's disease), the results of the French multicenter trial results (and of unknown trials) should be fully discussed before starting a new trial, including the choice of the infliximab dose. Intermediate safety rules should probably be scheduled.
References
- Corticosteroids improve short-term survival in patients with severe alcoholic hepatitis (AH): individual data analysis of the last three randomized placebo controlled double blind trials of corticosteroids in severe AH. J Hepatol. 2002;36:480–487
- . Cytokines in alcoholic and nonalcoholic steatohepatitis. N Engl J Med. 2000;343:1467–1476
- Combination of steroids with infliximab or placebo in severe alcoholic hepatitis: a randomized controlled pilot study. J Hepatol. 2002;37:448–456
- Anti-tumor necrosis factor-alpha monoclonal antibody therapy in severe alcoholic hepatitis. J Hepatol. 2003;38:419–425
- . Adrenocorticosteroid therapy and peptic-ulcer disease. N Engl J Med. 1984;310:201–202
- . Adrenocorticosteroid administration and peptic ulcer: a critical analysis. J Chronic Dis. 1985;38:457–468
- . Corticosteroids and peptic ulcer: meta-analysis of adverse events during steroid therapy. J Intern Med. 1994;236:619–632
- Survival and prognostic factors in patients with severe alcoholic hepatitis treated with prednisolone. Gastroenterology. 1996;110:1847–1853
- . Pentoxifylline improves short-term survival in severe acute alcoholic hepatitis: a double-blind, placebo-controlled trial. Gastroenterology. 2000;119:1637–1648
- Tuberculosis associated with infliximab, a tumor necrosis factor alpha-neutralizing agent. N Engl J Med. 2001;345:1098–1104
PII: S0168-8278(03)00045-X
doi:10.1016/S0168-8278(03)00045-X
© 2003 European Association for the Study of the Liver. Published by Elsevier Inc. All rights reserved.
