Alcoholic Hepatitis: Prognostic Models and Treatment




Severity scores for alcoholic hepatitis


Alcoholic hepatitis (AH) is a distinct subset of patients with alcoholic liver disease and has a potential for high mortality within 3 to 6 months after clinical presentation. Mild forms of AH usually improve with conservative management. However, patients with severe AH have been reported to have 30-day mortality of up to 50%. Therefore, assessment of the disease severity becomes an important and practical issue for clinicians involved in the management of patients with AH. Many scoring systems have been developed for use in clinical practice.


Discriminant Function Index


The discriminant function index (DFI) was initially described by Maddrey and colleagues in a placebo-controlled study to assess the benefit of corticosteroid (CS) therapy in 55 patients with AH. Using the formula: 4.6 × prothrombin time (PT) in seconds + serum bilirubin (mg/dL), patients with a DFI above 93 and treated with placebo had a 28-day survival of 25%, whereas those with a score of 93 or lower had 100% survival. In 1989, this score was modified (modified discriminant function or mDF) using prolongation of PT in seconds (over control) instead of absolute value of PT ( Table 1 ). Patients without treatment and mDF score of 32 or higher and/or the presence of encephalopathy had a 28-day survival of about 65%. A recent analysis confirmed this observation with untreated patients having 28-day survival of 68% among patients with mDF of 32 or higher. The American College of Gastroenterology recommends that AH patients with mDF score of 32 or higher should be considered for CS therapy.



Table 1

Scoring systems for assessment of severity of acute AH































































































Scoring System Parameters Formula Severe Disease
DFI SB and PT 4.6 × (patient’s PT – control PT in seconds) + SB ≥32
CPT SB, PT, serum albumin, ascites, and PSE 1 2 3 ≥7
SB <2 2–3 >3
PT↑ <4 4–6 >6
Albumin >3.5 2.8–3.5 <2.8
Ascites Absent Slight Tense
PSE None Grade I–II Grade III–IV
MELD score Age, SB, INR, and SC 9.57 log e (SC) + 3.78 log e (SB) + 11.2 log e (INR) + 6.43
Available at: www.mayoclinic.org/meld/mayomodel7.html 		≥21
GAHS Age, BUN, WBC, SB, and INR 1 2 3 ≥9
Age <50 ≥50 —
WBC <15 ≥15 —
BUN <14 ≥14 —
SB <7.3 7.3–14.6 >14.6
Tc>INR <1.5 1.5–2.0 >2.0
Lille score Age, labs at day 0 (SB, albumin, and PT), and change in SB at day 7 3.19–0.101 × age in yrs + 0.147 × albumin (g/L) on day 0 +0.0165 × change in SB (μmol/L) – 0.206 × RI (0 if absent and 1 if present) −0.0065 × SB on day 0 (μmol/L) −0.0096 × PT (in seconds) ≥0.45
ABIC score Age, bilirubin, INR, and creatinine (age × 0.1) + (SB × 0.08) + (SC × 0.3) + (INR × 0.8) ≥9

Patients with a CTP score of 5–6 have CTP stage A; 7–9, stage B; and 10–15, stage C.

Abbreviations: ABIC, Age, bilirubin, INR, creatinine; CTP, Child–Turcotte–Pugh score; DFI, discriminant function index; GAHS, Glasgow alcoholic hepatitis score; INR, International Normalized Ratio; MELD, Model for End-Stage Liver Disease; RI, renal insufficiency; PT, prothrombin time in seconds; SB, serum bilirubin in mg/dL; SC, serum creatinine in mg/dL; WBC, white blood count in 10 9 /L.


The advantages of the mDF are its simplicity of calculation and validation in many clinical trials. However, nonstandardization of the PT testing with laboratory to laboratory variation depending on the type of thromboplastin used by the laboratory is a limitation. Because patients with a DFI greater than 32 and an absence of encephalopathy are not considered for specific treatment, the mDF should have an accuracy of close to 100% in predicting their survival. However, patients with an mDF of less than 32 may have a 28-day mortality rates of about 7% to 17%. Overall sensitivity and specificity of predicting mortality in 1 study was 67% and 62%, respectively. One of the ways to tackle this issue is to lower the threshold score for initiating treatment. However, the risk benefit ratio does not favor CS treatment of patients with mDF score of less than 32.


Child–Turcotte–Pugh Score


This scoring system is based on 3 objective (serum bilirubin, serum albumin, and PT) and 2 subjective (hepatic encephalopathy and ascites) variables and categorize into 3 stages (A–C) with a total score of 5 to 6, 7 to 9, and greater than 9, respectively. The score is traditionally used for cirrhotic patients with mortality rates of about 10% to 15%, 25% to 30%, and 70% to 80% at 1 year for stages A, B, and C, respectively. Although not a traditional scoring system for AH patients, the Child–Turcot–Pugh (CTP) score was useful in predicting mortality at 3 to 6 months (see Table 1 ). Presently, the CTP score is not widely used for assessing severity of AH.


Model for End-Stage Liver Disease Score


Model for End-Stage Disease (MELD) score is being widely used for prediction of mortality in end-stage liver disease and is universally used to prioritize patients for liver transplantation (LT). Some studies have examined the use of MELD in assessing severity of AH. In a study on 34 patients with AH, the area under receiver operating characteristic curve (AUROC) for the MELD score was 0.82 with sensitivity and specificity in predicting 30-day mortality for a MELD score above 11 being 86% and 81%, respectively. In another study on 73 patients with AH, MELD score of 21 had highest sensitivity and specificity to predict mortality at 30 and 90 days. In yet another study on 202 patients, a first week MELD score of 20 or greater and an increase of MELD score at 1 week of 2 or more points had the highest sensitivity and specificity in predicting mortality. Recently, MELD-Na has been shown to be superior than MELD score among AH patients and ascites.


The advantage of MELD score is the use of the International Normalized Ratio (INR) (instead of PT). This is more comparable across laboratories, because the calculation accounts for the sensitivity of the thromboplastin reagent used in the test. However, the formula for calculating MELD is complex; however, this can be easily overcome using a computer program or calculating the MELD score online (see Table 1 ). Another problem with the use of MELD score is variation across studies in the best cutoff score in predicting mortality from 11, 18, to 21 to 22. Much of the variability, though, is due to use of original versus newer iterations of the MELD score that assign points differently. American Association for Study of Liver Diseases guidelines on the management of acute liver disease suggest that a cutoff MELD score of 18 be taken to predict severe AH and be the criterion for initiating treatment.


Glasgow Alcoholic Hepatitis Score


The Glasgow alcoholic hepatitis score (GAHS) was introduced in 2005 to assess AH severity (see Table 1 ). The score ranges between 5 and 12; a GAHS of 9 or higher at days 1 and 7 was more accurate than the mDF in predicting survival at 28 days ( P = .0016 and P = .0038, respectively) and at 84 days ( P = .0179 and P = .0477, respectively). GAHS at day 7 but not at day 1 was better in predicting 28-day outcome than the MELD score ( P = .0339 and P = .069). GAHS at days 1 and 7 was more accurate than the MELD score in predicting the 84-day outcome ( P = .0005 for both scores). GAHS was reliable in predicting mortality irrespective of the use of INR or PT for determination of coagulation status or use of liver biopsy for diagnosis of AH. GAHS was confirmed in a validation cohort of 195 patients in this study; however, the score has not been validated outside the United Kingdom.


Recently, the GAHS has been shown to predict response to CS treatment. In this retrospective study, 144 AH patients with an mDF of 32 or higher were included. Of this group, 73 (51%) with a GAHS of 9 or greater were treated with CS. Patients with a GAHS below 9 did not differ on 28- and 84-day survival whether or not treated with CS (84% vs 80% and 73% vs 68%, respectively). However, the survival of patients with a GAHS of 9 or greater was better with CS treatment at day 28 (78% vs 52%; P = .002) and at day 84 (59% vs 38%; P = .02).


Age–Bilirubin–INR–Creatinine Score


A new score, the age–bilirubin–INR–creatinine (ABIC) score, has been suggested for prognostic stratification of patients with AH (see Table 1 ). Using the cutoff values of 6.71 and 9.0, authors identified patients with low, intermediate, and high risk of death at 90 days (100%, 70%, and 25% of survival rate, respectively; P <.0001). Using the same cutoff values, the ABIC score also stratified patients according to their risk of death at 1 year. The AUROC for predicting mortality at 90 days using ABIC score was 0.82 (95% confidence interval [CI], 0.73–0.91; P = .0001). In comparison, AUROC values were lower for the mDF score, MELD score, and GAHS: 0.70 (95% CI, 0.56–0.84; P = .008), 0.76 (95% CI, 0.64–0.88; P = .0004), and 0.75 (95% CI, 0.63–0.86; P = .001), respectively. On multivariate analysis, ABIC score was the best independent predictor of 90-day mortality (hazard ratio, 2.78; 95% CI, 1.90–4.09; P = .0001).


Early Change in Bilirubin Level


In a retrospective study on 238 biopsy-proven AH patients, a decrease in serum bilirubin at 1 week or early change in bilirubin level (ECBL) was a predictor for survival. A total of 73% patients showed ECBL at 1 week (4.9–4.4) and 83% of them survived. In contrast, the remaining cases did not show ECBL, with only 23% survival. In another study reported from France assessing role of N -acetylcysteine (NAC) in combination with CS, decrease in bilirubin at day 14 was a significant predictor of survival.


Lille Model


About 40% of patients with severe AH fail to respond to treatment with steroids. In a prospective study on 320 biopsy-proven severe AH, nonresponders to steroids (NRS) could be identified based on ECBL and other five variables. This led to the development of Lille score (see Table 1 ).


Survival at 6 months was lower for patients with a Lille score of 0.45 or higher compared with patients with Lille score of less than 0.45 (25% vs 85%; P <.0001). The AUROC value for the Lille score cutoff of 0.45 was higher than the CTP score (0.89 vs 0.62; P <.00001) or mDF score (0.89 vs 0.66; P <.00001). The authors concluded that CS be discontinued for patients with a Lille score of 0.45 or greater at 1 week. The Lille score maintains accuracy in predicting the survival when used across a range. In a retrospective study on 641 biopsy-proven AH, a linear correlation with survival was seen among groups with Lille score of less than 0.16, 0.16 to 0.56, or greater than 0.56 with survival rates of 87%, 70%, and 21%, respectively, at 6 months. Although this score has not been validated outside France, the score has been validated prospectively in other studies from the same center. Unfortunately, the score does not guide initiation of treatment because it cannot be calculated at admission.


Comparison of Scores


Many studies have compared available scoring systems assessing severity of AH. MELD and DFI have been compared among 6 studies. The data have shown differences across studies ( Table 2 ).


Tags:
Sep 7, 2017 | Posted by in GASTROENTEROLOGY | Comments Off on Alcoholic Hepatitis: Prognostic Models and Treatment

Full access? Get Clinical Tree
Get Clinical Tree app for offline access