Steroids for Severe Alcoholic Hepatitis: More Risk Than Reward?

Watch a video presentation of this article

Abbreviations

AH

alcoholic hepatitis

bDNA

bacterial DNA

CI

confidence interval

LT

liver transplantation

OR

odds ratio

SAH

severe alcoholic hepatitis

STOPAH

Steroids or Pentoxifylline for Alcoholic Hepatitis

Alcoholic hepatitis (AH) is a condition marked by inflammatory liver injury in patients with chronic excessive alcohol consumption. Severe alcoholic hepatitis (SAH), defined by a Maddrey’s Discriminant Function greater than 32, can be associated with greater than 70% mortality rate within 6 months in patients who demonstrate persistent liver impairment even after medical therapy.1, 2 Historically, the primary treatment option for SAH has been corticosteroids, although reported efficacy has been variable.

Prior studies have suggested that corticosteroids may improve short‐term survival in patients with SAH.3, 4 However, patients with SAH who experience development of infection after corticosteroid treatment have a higher rate of mortality.5 In 2015, Thursz et al.6 published results from the STOPAH (Steroids or Pentoxifylline for Alcoholic Hepatitis) trial, a multicenter study in which more than 1000 patients with SAH were randomized to 28 days of treatment with prednisolone, pentoxifylline, prednisone + pentoxifylline, or placebo. When specifically comparing patients with and without steroid exposure, the investigators identified a trend toward reduced 28‐day mortality in the steroid group, but this survival benefit was no longer seen at 90 days (Fig. 1). Notably, the rate of infection was nearly twice as high in the group of patients who received corticosteroids, although mortality rates were similar across all groups.

Figure 1.

Prednisolone therapy is associated with a trend toward improved survival at 28 days, but not at 90 days, based on data from the STOPAH trial.

Based on these results, Vergis et al.7 recently analyzed data from the original STOPAH cohort to elucidate the relationship between infection and clinical outcomes in patients with SAH treated with and without corticosteroids and to further determine whether circulating serum levels of 16S ribosomal bacterial DNA (bDNA) could predict development of infection in patients undergoing medical therapy for SAH. In total, 1092 patients with SAH from the original STOPAH trial were included in this analysis, with 547 patients having been treated with prednisolone. Patients were enrolled at multiple sites in the United Kingdom from 2011 to 2014. The study used a double‐blind, 2×2 design with randomization of patients to one of four 28‐day treatment regimens: (1) prednisolone 40 mg daily, (2) pentoxifylline 400 mg three times a day, (3) neither medication, or (4) both medications. The investigators specifically evaluated data regarding infection and mortality. Whole blood samples were obtained from a subset of patients to measure bDNA levels by polymerase chain reaction.

Among all study patients, 23% experienced development of infections on treatment, whereas 10% had infections posttreatment. Median time to development of infection was 13 days after the start of treatment. Notably, higher baseline discriminant function and Model for End‐Stage Liver Disease score were associated with increased risk for infection.

Prednisolone therapy was associated with an increased risk for posttreatment infections (10% versus 6%; odds ratio [OR] 1.70; 95% confidence interval [CI]: 1.07‐2.69) but not on‐treatment infections (Fig. 2). Specifically, patients exposed to steroids were more likely to experience infections classified as serious. Pentoxifylline use did not correlate with higher risk for on‐treatment or posttreatment infections. In patients treated with prednisolone, incident infection was significantly associated with mortality (39% versus 22%; OR 2.27; 95% CI: 1.52‐3.38). In contrast, there was no significant association between development of infection and mortality in the patients who did not receive corticosteroids. Among patients treated with prednisolone, failure to demonstrate early improvement (i.e., Lille score >0.45) was associated with a higher risk for development of infection. Development of early infection within 7 days of prednisolone initiation correlated with increased 90‐day mortality.

Figure 2.

Treatment with prednisolone is associated with an increased rate of posttreatment, but not on‐treatment, infections.

Among the 731 patients with SAH with baseline blood samples for bDNA analysis, 90% had detectable bDNA levels. When patients treated with prednisolone were analyzed, higher baseline bDNA levels were associated with development of infection within 7 days of treatment initiation. No such association between bDNA levels and early infection risk was found among patients not treated with corticosteroids. Finally, avoiding prednisolone therapy in patients with high bDNA levels was associated with reduced 90‐day mortality.

Although well‐designed, this study does suffer from two main limitations. First, the lack of standardized liver biopsy in this protocol means that many patients were diagnosed with AH based on clinical criteria. Second, only 40% of infection diagnoses were based on positive culture data. It is conceivable that infection may have been overdiagnosed in patients with SAH who frequently exhibit features of inflammatory response such as leukocytosis.

Results from the original STOPAH trial suggest that prednisolone therapy does not significantly improve 90‐day survival and is associated with a higher infection rate in treated patients. The findings from the study by Vergis et al.7 further demonstrate that (1) patients with SAH are inherently at higher risk for infection; (2) prednisolone therapy may increase the risk for development of infection even after steroid therapy is complete; and (3) prednisolone use may independently increase the risk for mortality in patients who experience infection while taking or after a course of prednisolone.

Although development of the Lille score has allowed clinicians to limit the duration of steroid therapy in nonresponders to as little as 4 days, the main limitation of the Lille model is the need to wait for a day 4 bilirubin level.8 Results from the study by Vergis et al.7 suggest that corticosteroid nonresponders are already at higher risk for development of infection, and development of infection within 7 days of prednisolone initiation increases the risk for subsequent 90‐day mortality. As such, there would be utility in having an early prognostic marker similar to bDNA to better stratify the risks versus benefits of initiating corticosteroids for individual patients.

In the absence of effective medical therapy, there has been increasing interest in the use of liver transplantation (LT) for carefully selected patients with SAH. Short‐term survival after LT has been acceptable in European and US experiences, but early posttransplant bacterial and fungal infections have been observed.9, 10 A trial of corticosteroids before transplantation has failed in many of these patients, which would suggest that steroid exposure is a likely risk factor for development of early posttransplant infections. Furthermore, the more recent US multicenter cohort demonstrated a correlation between pretransplant steroid exposure and increased mortality because of early posttransplant infection. Therefore, until biomarkers for earlier prediction of steroid response are identified or alternative medical therapies are discovered, steroid avoidance may be the best approach in patients with SAH.