The Management of Alcoholic Hepatitis
The Management of Alcoholic Hepatitis
Background The assessment of alcoholic hepatitis remains controversial. Several scores have been developed or used for this purpose.
Aim To study the use of the Glasgow Alcoholic Hepatitis Score (GAHS), the Discriminant Function (DF), Model for End-Stage Liver Disease (MELD) and the ABIC (age, bilirubin, INR and creatinine) scores as well as scores to assess corticosteroid response in the management of alcoholic hepatitis.
Methods A total of 182 patients were studied prospectively. The GAHS, MELD, ABIC and DF scores were recorded on admission and serially over the first week of hospital management. Treatment with corticosteroids or pentoxifylline was considered if the GAHS was ≥9.
Results There were no differences in outcome between favourable scores as per recommended cut-off points. Patients with a GAHS<9 had similar outcome whether their MELD, DF or ABIC scores were favourable or unfavourable. Treated patients with a GAHS≥9 had a significantly better 90-day outcome than those who did not: 58% and 30% respectively, P = 0.01; HR 0.33 (0.14, 0.78).
Patients treated with corticosteroids who had a fall in bilirubin of 25% after a week of treatment had an improved survival: 82% compared with 44% [P = 0.0005: HR 3.70 (1.77, 7.73)]. The Lille Score or a 25% fall in bilirubin had greater sensitivities than an early change in bilirubin level (95% and 90% compared with 58%) to assess treatment response.
Conclusions In this single-centre study, a GAHS ≥9 identified patients who may benefit from treatment of alcoholic hepatitis. Intention-to-treat randomised-controlled trials using a GAHS ≥9 as the threshold for treatment are needed to validate these findings. Response to corticosteroids can be assessed using the Lille Score or by a 25% fall in bilirubin.
Alcoholic hepatitis remains a life-threatening manifestation of alcoholic liver disease seen frequently in clinical practice. In its more severe forms, it carries a short-term mortality of up to 60%. Despite the prevalence of this condition, controversy persists regarding its assessment. Originally described in 1978 and then modified in 1989, the discriminant function (DF) was the first disease-specific score for alcoholic hepatitis. Whilst the DF was a ground-breaking advance, recently, doubts have been raised regarding its reproducibility and accuracy.
The Glasgow Alcoholic Hepatitis Score (GAHS) has been developed as a disease-specific score for alcoholic hepatitis. The GAHS was derived from a population of 241 patients and validated in a population of 195 patients from throughout the United Kingdom. In its original description, it was found to be statistically superior to the DF. Further data provided evidence that patients with a GAHS ≥9 may benefit from treatment with corticosteroids, whereas those with a GAHS <9 are unlikely to benefit from such treatment even if the DF is ≥32.
The Model for End-Stage Liver Disease (MELD) and the ABIC (age, bilirubin, INR and creatinine) scores have also been proposed for the assessment of alcoholic hepatitis. The MELD, however, has never been shown to be statistically superior to the DF. In addition, the threshold, or optimal cut-point, of MELD score for identifying patients with a poor prognosis varies widely between 11 and 30.5, depending upon the timing of the score calculation, the version of MELD score employed and the end-point studied. However, a value of 18 using the UNOS modification of the MELD has been proposed by AASLD as the threshold to consider initiating treatment. The ABIC Score alternatively uses the variables such as age, serum bilirubin, INR and serum creatinine to stratify disease severity into low, intermediate and high risk of death. However, even those with a 'low' risk of death may have a mortality of over 17% after 84 days.
Whilst there may be little to chose between these scores when it comes to accuracy as assessed by Area under the Receiver Operating Characteristic (AUROC) curve analysis, the question remains as to which is most useful in clinical practice. An effective score has to not only identify patients with a poor prognosis but also guide treatment decisions to improve patient outcomes.
Another area of discussion is whether it is possible to identify patients who will have a good response to corticosteroids. Initially, it was suggested that any fall in serum bilirubin 1 week after starting treatment was an indication of corticosteroid responsiveness (Early Change in Bilirubin Level: ECBL). This concept was taken further with a proposal that a 25% fall in bilirubin from baseline after 1 week of corticosteroid treatment was a more reproducible indicator of response. More recently, the Lille prognostic model has been developed as a more accurate predictor of outcome for corticosteroid-treated patients. Since this initial description, a further sub-stratification of complete, partial and nonresponders has been proposed.
Therefore, there is debate on assessment of both the initial presentation and subsequent response to treatment of alcoholic hepatitis. In this context, we prospectively assessed all alcoholic hepatitis patients attending Glasgow Royal Infirmary where we have been using the GAHS to assess disease severity and to guide treatment decisions.
Abstract and Introduction
Abstract
Background The assessment of alcoholic hepatitis remains controversial. Several scores have been developed or used for this purpose.
Aim To study the use of the Glasgow Alcoholic Hepatitis Score (GAHS), the Discriminant Function (DF), Model for End-Stage Liver Disease (MELD) and the ABIC (age, bilirubin, INR and creatinine) scores as well as scores to assess corticosteroid response in the management of alcoholic hepatitis.
Methods A total of 182 patients were studied prospectively. The GAHS, MELD, ABIC and DF scores were recorded on admission and serially over the first week of hospital management. Treatment with corticosteroids or pentoxifylline was considered if the GAHS was ≥9.
Results There were no differences in outcome between favourable scores as per recommended cut-off points. Patients with a GAHS<9 had similar outcome whether their MELD, DF or ABIC scores were favourable or unfavourable. Treated patients with a GAHS≥9 had a significantly better 90-day outcome than those who did not: 58% and 30% respectively, P = 0.01; HR 0.33 (0.14, 0.78).
Patients treated with corticosteroids who had a fall in bilirubin of 25% after a week of treatment had an improved survival: 82% compared with 44% [P = 0.0005: HR 3.70 (1.77, 7.73)]. The Lille Score or a 25% fall in bilirubin had greater sensitivities than an early change in bilirubin level (95% and 90% compared with 58%) to assess treatment response.
Conclusions In this single-centre study, a GAHS ≥9 identified patients who may benefit from treatment of alcoholic hepatitis. Intention-to-treat randomised-controlled trials using a GAHS ≥9 as the threshold for treatment are needed to validate these findings. Response to corticosteroids can be assessed using the Lille Score or by a 25% fall in bilirubin.
Introduction
Alcoholic hepatitis remains a life-threatening manifestation of alcoholic liver disease seen frequently in clinical practice. In its more severe forms, it carries a short-term mortality of up to 60%. Despite the prevalence of this condition, controversy persists regarding its assessment. Originally described in 1978 and then modified in 1989, the discriminant function (DF) was the first disease-specific score for alcoholic hepatitis. Whilst the DF was a ground-breaking advance, recently, doubts have been raised regarding its reproducibility and accuracy.
The Glasgow Alcoholic Hepatitis Score (GAHS) has been developed as a disease-specific score for alcoholic hepatitis. The GAHS was derived from a population of 241 patients and validated in a population of 195 patients from throughout the United Kingdom. In its original description, it was found to be statistically superior to the DF. Further data provided evidence that patients with a GAHS ≥9 may benefit from treatment with corticosteroids, whereas those with a GAHS <9 are unlikely to benefit from such treatment even if the DF is ≥32.
The Model for End-Stage Liver Disease (MELD) and the ABIC (age, bilirubin, INR and creatinine) scores have also been proposed for the assessment of alcoholic hepatitis. The MELD, however, has never been shown to be statistically superior to the DF. In addition, the threshold, or optimal cut-point, of MELD score for identifying patients with a poor prognosis varies widely between 11 and 30.5, depending upon the timing of the score calculation, the version of MELD score employed and the end-point studied. However, a value of 18 using the UNOS modification of the MELD has been proposed by AASLD as the threshold to consider initiating treatment. The ABIC Score alternatively uses the variables such as age, serum bilirubin, INR and serum creatinine to stratify disease severity into low, intermediate and high risk of death. However, even those with a 'low' risk of death may have a mortality of over 17% after 84 days.
Whilst there may be little to chose between these scores when it comes to accuracy as assessed by Area under the Receiver Operating Characteristic (AUROC) curve analysis, the question remains as to which is most useful in clinical practice. An effective score has to not only identify patients with a poor prognosis but also guide treatment decisions to improve patient outcomes.
Another area of discussion is whether it is possible to identify patients who will have a good response to corticosteroids. Initially, it was suggested that any fall in serum bilirubin 1 week after starting treatment was an indication of corticosteroid responsiveness (Early Change in Bilirubin Level: ECBL). This concept was taken further with a proposal that a 25% fall in bilirubin from baseline after 1 week of corticosteroid treatment was a more reproducible indicator of response. More recently, the Lille prognostic model has been developed as a more accurate predictor of outcome for corticosteroid-treated patients. Since this initial description, a further sub-stratification of complete, partial and nonresponders has been proposed.
Therefore, there is debate on assessment of both the initial presentation and subsequent response to treatment of alcoholic hepatitis. In this context, we prospectively assessed all alcoholic hepatitis patients attending Glasgow Royal Infirmary where we have been using the GAHS to assess disease severity and to guide treatment decisions.
Source...