Characteristics of alcoholic liver disease and predictive factors for mortality of patients with alcoholic cirrhosis
Yan-Di Xie, Bo Feng, Yan Gao and Lai Wei
Beijing, China
AuthorAffiliations: Peking University People's Hospital, Peking University Hepatology Institute, Beijing Key Laboratory of Hepatitis C and Immunotherapy for Liver Disease, Beijing 100044, China (Xie YD, Feng B, Gao Y and Wei L)
CorrespondingAuthor: Lai Wei, MD, PhD, Peking University People's Hospital, Peking University Hepatology Institute, Beijing Key Laboratory of Hepatitis C and Immunotherapy for Liver Disease, Beijing 100044, China (Tel/Fax: 86-10-88325980; Email: weilai@pkuph.edu.cn)
© 2013, Hepatobiliary Pancreat Dis Int. All rights reserved.
doi: 10.1016/S1499-3872(13)60094-6
Contributors: WL proposed the study. XYD and FB performed research and wrote the first draft. XYD and GY collected and analyzed the data. All authors contributed to the design and interpretation of the study and to further drafts. WL is the guarantor.
Funding: None.
Ethical approval: Not needed.
Competing interest: No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
BACKGROUND: Alcoholic liver disease is one of the major chronic liver diseases worldwide. The aim of the study was to describe the clinical characteristics of alcoholic liver disease and to compare the predictive values of biochemical parameters, complications, Child-Turcotte-Pugh score, model for end-stage liver disease (MELD) score and discriminant function score for the mortality of in-hospital or 3-month after discharge of patients with alcoholic cirrhosis (AC).
METHODS: A retrospective record review and statistical analysis were performed on 205 consecutive patients with the discharge diagnosis of alcoholic liver disease. Three models were used to predict the mortality of patients with AC. The number of variceal hemorrhage, infection, hepatic encephalopathy and hepatocellular carcinoma was analyzed as "numbers of complications". Model 1 consisted of creatinine, white blood cell count, international normalized ratio and "numbers of complications". Model 2 consisted of MELD score. Model 3 included "numbers of complications" and MELD score.
RESULTS: The risk of developing AC was significant for patients with alcohol consumption of higher than 80 g/d (OR=2.807, P<0.050) and drinking duration of longer than 10 years (OR=3.429, P<0.028). The area under curve for predicting in-hospital mortality of models 1, 2 and 3 was 0.950, 0.886 and 0.911 (all P<0.001), respectively. The area under curve for predicting the 3-month mortality of models 1, 2 and 3 was 0.867, 0.878 and 0.893 (all P<0.001), respectively.
CONCLUSIONS: There is a dose-dependent relationship between alcohol consumption and the risk of developing AC. MELD score has a better predictive value than Child-Turcotte-Pugh or discriminant function score for patients with AC, and model 1 or 3 is better than model 2.
(Hepatobiliary Pancreat Dis Int 2013;12:594-601)
KEY WORDS: alcoholic liver disease; model for end-stage liver disease; mortality; prognostic factors; risk factors
Introduction
Alcoholic liver disease (ALD) is one of the major chronic liver diseases worldwide. The clinical and histological phases of ALD include alcoholic fatty liver (AFL), alcoholic hepatitis (AH) and alcoholic cirrhosis (AC). Although 90%-95% of heavy drinkers suffer from AFL, only 20%-40% of them develop AH and 8%-20% eventually progress to AC.[1] Several risk factors for the development of AC have been identified. The amount and duration of alcohol consumption are the two important factors and the risk thresholds vary widely among different races.
Patients with AC have a poor prognosis. Follow-up studies showed that the 1- and 5-year mortalities are around 30% and 60%, respectively.[2] Therefore, it is important to recognize AC by using laboratory tests and imaging techniques, and to establish accurate prognostic models for these patients because of adequate therapies including transplantation.
The clinical and biochemical parameters that have been used to accurately predict the prognosis of cirrhotic patients include age, the amount of alcohol consumption, the aspartate transaminase to alanine transaminase ratio (AST/ALT ratio),[3] histological severity, creatinine (CRE),[4] white blood cell (WBC) count,[5] and complications such as ascites,[4] hepatic encephalopathy (HE),[6] variceal bleeding[7] and spontaneous bacterial peritonitis (SBP).[8]
Several prognostic models have been developed so far. The Child-Turcotte-Pugh (CTP) score is still widely used in the evaluation of prognosis of cirrhotic patients, although it has some drawbacks such as the subjectivity of clinical parameters and a limited discriminatory ability.[9] The discriminant function (DF) score is used most often in the prognosis of AH for the prediction of possible benefit of corticosteroids therapy.[10] There are no matured clinical parameters and/or system to predict the prognosis of AC. Model for end-stage liver disease (MELD) score has been proven to be valuable under different settings for patients with cirrhosis.[9]
In this study, we aimed to describe the clinical characteristics of ALD, to find out the relationship between alcohol intake and the development of AC in the Chinese population, and to compare the predictive values of biochemical parameters, complications, MELD, CTP, and DF scores for the in-hospital or 3-month mortality of patients with AC.
Methods
Patients
A total of 205 consecutive in-patients with ALD who had been treated at the Liver Department of Peking University People's Hospital from January 2000 to December 2011 were enrolled in this study.
The clinical diagnosis criteria of ALD[12] were (a) a history of excess alcohol consumption of ≥40 g/d for men and ≥20 g/d for women over 5 years; (b) the presence of clinical, biochemical and/or detectable morphology abnormalities suggesting liver injury; (c) negative serology for hepatitis B surface antigen (HBsAg) and hepatitis C antibody (anti-HCV); (d) the absence of other causes of chronic liver disease.
The clinical classification criteria of ALD[12] were (a) AFL: detection of fatty liver using imaging techniques, negative of obesity, and slightly or no elevation in liver function test parameters; (b) AH: the presence of one or more of the following: elevated levels of serum ALT, AST, alkaline phosphatase (ALP), gamma-glutamyl transpeptidase (GGT), total bilirubin (TBIL), and no evidence of cirrhosis; (c) AC: cirrhosis confirmed by laboratory findings, radiographic evidence or histology.
Methods
Data were retrieved from hospital discharge record. A total number of 391 consecutive patients with ALD were reviewed, 177 patients were excluded because of the positive serology for HBsAg or anti-HCV, 5 were due to drug-induced liver injury and 4 were due to incomplete data. 205 patients met the inclusion criteria (Fig. 1).
The demographic data included age, gender, nationality, history of alcohol consumption, cigarette smoking and accompanied diseases. The clinical data included the results of laboratory tests, reports of imaging techniques, and complications such as ascites, HE, variceal hemorrhage, infection, hepatorenal syndrome (HRS), and/or hepatocellular carcinoma (HCC). Types of infection included SBP, pneumonia, urinary infection, cellulitis or sepsis. The CTP, MELD and DF scores were calculated. In-hospital death was the primary outcome. Follow-ups by telephone or records review were conducted for all patients to check 3-month survivals after discharge.
The diagnosis of SBP was based on the analysis of ascitic fluid with a neutrophil count ≥0.25×109/L. The diagnosis of HRS was based on the definition set forth by the International Ascites Club, which was (a) the CRE concentration of above 1.5 mg/dL; (b) the absence of any other causes for renal disease; (c) urine sodium concentration <10 mmol/L off diuretics; (d) the urine osmolality of greater than that of plasma; and (e) the lack of improvement in renal function after volume expansion and if pertinent after withdrawal of diuretics.[13] The MELD score formula was 3.8×loge (TBIL [mg/dL])+11.2×loge(INR)+9.6×loge(CRE[mg/dL])+6.4×(etiology: 0 if cholestatic or alcoholic, 1 otherwise).[9] In our study, the constant 6.4 was not used because the etiology of liver disease was alcoholic. The CTP score was according to the method described by Pugh et al.[14] The DF score formula was 4.6×(patient's INR-control INR)+TBIL [mg/dL].[10]
The cause of death was separated into four groups: (a) renal insufficiency; (b) variceal hemorrhage; (c) infection with resultant sepsis; and (d) multiple organ failure, excluding those directly related to infection or HRS. Sepsis was defined as the presence of infection, bacteremia, hemodynamic instability, and systemic inflammatory response syndrome.
Statistical analysis
If continuous variables were normally distributed, data were presented as mean with standard deviation (mean±SD). When they were not normally distributed, median with quartile range (median±QR) was presented, and non-parametric test was used to test differences. The Chi-square test was used to compare ratios.
To analyze risk factors for in-hospital and 3-month mortality, the logistic regression analysis was used. Individual variables with P value of less than 0.2 were included in the univariate analysis first. Then variables with P value of less than 0.05 entered the second multivariate analysis model. Odds ratio (OR) and 95% confidence interval (CI) were also calculated.
To assess the prognostic utilities of the models, receiver operating characteristic (ROC) curves were plotted. The area under curve (AUC) was computed by the Statistical Package for Social Sciences (SPSS, version 16.0; Chicago, IL, USA) statistical program. An AUC over 0.7 was considered clinically useful, and that between 0.8 and 0.9 indicated an excellent diagnostic accuracy. An AUC of over 0.9 was seldom seen.
All analyses were performed using the SPSS. A P value of <0.05 was considered statistically significant.
Results
Demographics and clinical characteristics
The 205 patients met the inclusion criteria for the analysis. Among them, 13 (6.3%), 33 (16.1%) and 159 patients (77.6%) fulfilled the clinical criteria for AFL, AH and AC, respectively. The demographics and clinical characteristics of the 205 patients are shown in Table 1.
The biochemical data were compared among the AFL, AH and AC groups (Table 1). ALT, AST and GGT were the highest in the AH group. The AST/ALT ratio of the AFL group (0.75±0.43) was significantly lower than that of the AH group (1.15±1.75), and the AST/ALT ratio of the AH group AH was significantly lower than that of the AC group (2.00±1.20). The ratio was ≤1 in 76.9%, 45.5% and 7.5% of the patients in the AFL, AH and AC groups (P<0.001), respectively. The ratio was ≥2 in 0%, 33.3% and 53.5% of the patients in the AFL, AH and AC groups (P<0.001), respectively (Fig. 2).
Of the 205 patients, their chief complaints were fatigue (56.6%), abdominal distension (55.1%), poor appetite (34.1%), leg edema (41.0%) and jaundice (35.1%). One hundred and twenty-four (60.5%) patients presented with ascites. This was followed by infection (34.6%), variceal hemorrhage (14.6%), HE (14.6%), and HCC (3.0%).
There were 14 deaths during the hospitalization, accounting for a mortality of 6.8%. Seven patients died from renal insufficiency, 5 from variceal hemorrhage, and 2 from multi-organ failure and sepsis. Fourteen patients were lost to follow-up, and 9 died within 3 months after discharge. The survival and in-hospital death variables and scores are shown in Table 2. The CRE, INR, TBIL, CTP, MELD and DF scores were significantly higher in the death group (P<0.05).
Alcohol consumption and AC
Table 3 shows the distribution of alcoholic cirrhosis and alcohol consumption. The proportion of AC increased with the amount of daily alcohol consumption and the duration of drinking. The risk to AC was significant with an alcohol consumption of more than 80 g/d (OR=2.807, P<0.050) or a duration of longer than 10 years (OR=3.429, P<0.028) (data not shown). The highest OR (5.244, P<0.010) was seen when the amount of daily drinking exceeded 200 g. At this level of alcohol consumption, the percentage of individuals with AC was 86.6%, which was significantly higher than that in patients with alcohol consumption of less than 80 g/d.
Risk factors for in-hospital mortality of patients with cirrhosis
A total of 41 variables of the AC group were chosen for logistic regression analysis, including demography, alcohol consumption, laboratory tests, complications, CTP, MELD and DF scores. The results of univariate analysis are shown in Table 4. Age, TBIL, CRE, WBC count, INR, the presence of variceal hemorrhage, infection, HE and HCC, CTP, MELD and DF scores were associated with in-hospital mortality. Variables such as TBIL, CRE, INR and the presence of HE were components used for calculating different scores. Hence variables were classified into two models. The number of variceal hemorrhage, infection, HE and HCC were analyzed as "numbers of complications", coded from 0 to 4. Multivariate analyses of six variables (model 1) showed that CRE, WBC count, INR and "numbers of complications" were associated with mortality. Similarly, multivariate analyses of different scores (model 2) were performed. MELD score (OR=1.283, P<0.001) was associated with mortality. The results of univariate analysis showed that CTP and DF scores were associated with an increased risk of death. However, according to the results of multivariate analysis, neither CTP nor DF score accurately predicted the mortality.
To further evaluate the prognostic value of the two models, ROC curves were graphed and AUCs were computed (Fig. 3A). The AUCs for models 1 and 2 were 0.950 and 0.886 (P<0.001), respectively.
Complications such as ascites, HE, variceal bleeding and SBP are important prognostic factors. As mentioned above, AUC of model 1 including "numbers of complications" was superior to MELD score alone. Therefore, we combined "numbers of complications" with the MELD score, which was named model 3. ROC curve was graphed and the AUC for model 3 was 0.911 (P<0.001) (Fig. 3A).
Risk factors for 3-month mortality of patients with cirrhosis
In this analysis, we used the same variables and statistical analysis methods as those in the analysis of risk factors for the in-hospital mortality of patients with AC. The results are shown in Table 5. In model 1, CRE, WBC count, INR, and "numbers of complications" were associated with the mortality. In model 2, MELD score (OR=1.318, P<0.001) was associated with the mortality. CTP and DF scores were independently associated with the 3-month mortality.
The AUC for models 1, 2 and 3 was 0.867, 0.878 and 0.893 (all P<0.001), respectively (Fig. 3B).
Discussion
A dose-dependent relationship between alcohol consumption and the risk of developing AC was observed in this study. The risk of developing AC was significant with an alcohol consumption of more than 80 g/d or longer than 10 years. As mentioned in Mandayam's study, the risk of developing cirrhosis increases with an alcohol consumption of more than 60-80 g/d for 10 years in men.[15] There is no consensus on the relationship between the development of AC and the risk threshold of daily alcohol consumption from 24 g[16] to 120 g.[17] Besides alcohol consumption, there are other risk factors for the development of AC such as gender, obesity, cigarette smoking and infection with chronic hepatitis viruses. As reported, elevated level of AST is observed in all forms of ALD with a sensitivity of 50% and a specificity of 80%.[18] We found the elevated levels of AST and GGT in all groups and the highest in AH patients. In contrast, AC was suspected if albumin was decreased, INR was prolonged or thrombocytopenia was seen. There was a gradual but significant increase in the AST/ALT ratio from patients with AFL to those with AC. The ratio was ≥2 in 53.5% of the patients with AC (P<0.001). The mean and maximum ratio in patients with AC was 2.3 and 20.8, respectively. Nyblom and colleagues[3] found that the mean and maximum ratio of AST/ALT in patients with alcoholic cirrhosis was 2.6 and 12.1, respectively. Nyblom's study showed that a high AST/ALT ratio indicated an increase in severity of alcoholic liver injury.
Patients with cirrhosis have a poor prognosis. It is important to find an accurate survival predictive system for these patients. In this study, all three models had good discrimination in predicting in-hospital and 3-month mortality in cirrhotic patients. All three models predicted the in-hospital mortality (AUCs of 0.950, 0.886 and 0.911) slightly better than the 3-month mortality (AUCs of 0.867, 0.878 and 0.893). Our results are consistent with those of the study by Kamath,[9] demonstrating that the AUC of MELD score used for predicting the 1-week and 3-month mortality was 0.95 and 0.87, respectively.
The components of models determine the prognostic values in AC. (a) Both CRE and INR, indices of renal function and a measure of coagulation system function, are well-recognized as predictors of mortality in patients with liver diseases.[4, 19] (b) Significant correlation was found between WBC count and infection. Infection occurs in patients with advanced cirrhosis and is associated with a bad prognosis.[5] (c) Bilirubin, reflecting liver function, is a poor prognostic factor in patients with ALD.[20] Although TBIL is included in MELD score, it is not an independent predictor of death in the multivariable analysis. This might be one of the reasons why MELD score had a lower predictive value than the other two models in our study. (d) Complications such as ascites,[4] encephalopathy,[6] variceal bleeding[7] and SBP[8] adversely affect the survival of patients with cirrhosis. However, "numbers of complications" combined with MELD score (model 3) had minimal effect in improving the prediction of in-hospital and 3-month mortality, and the increase in AUC was only 0.025 and 0.015, respectively. Similar result has been found in a previous study.[9]
In terms of predicting in-hospital or 3-month mortality, the values of AUCs of both models 1 and 3 were higher than those of MELD scores. Specifically, the AUCs of models 1 and 3 were over 0.9, being excellent in predicting in-hospital mortality. However, a disease severity model should not only have a sound statistical result, but also use available and objective determinants to easily calculate and generalize a heterogeneous group of patients.[21] Therefore, further prospective study is needed to validate the prognostic utility of models 1 and 3 for AC.
In our study, AFL, AH and AC were defined on clinical parameters. Because liver biopsy is an invasive procedure, it is not recommended for all patients with suspected ALD but clinical diagnosis of ALD is acceptable.[12] In China, women drinkers are few, and their levels of alcohol consumption are low, thus few women suffer from ALD. In our study, 204 of the 205 patients were male. We therefore did not analyze the relationship between gender and ALD.
Malinchoc et al[11] developed and validated MELD score for predicting 3-month survival after transjugular intrahepatic portosystemic shunt (TIPS). Angermayr and colleagues[22] also compared the accuracy of MELD and CTP scores in predicting the 3-month survival after TIPS. Kim et al[23] studied patients with liver cirrhosis who were on a waiting list for liver transplantation, and developed and validated the MELD-sodium model for predicting the 3-month mortality after registration. In this study, we chose a period of 3 months because the clinical scenario might vary widely and beyond the period during which other factors would affect the analysis. Furthermore, the average waiting times for liver transplantation for patients with a United Network for Organ Sharing status of 2 are approximately 3 months.[24]
There were some limitations in our study. This was a retrospective and single center study with a small sample size. Through this retrospective study we have found three valuable models, and we are designing a prospective study to validate these three models.
In conclusion, there is a dose-dependent relationship between alcohol consumption and the risk of developing AC. A high AST/ALT ratio indicates increased severity of alcoholic liver injury. MELD is a better model compared with CTP and DF in predicting the outcomes of patients with AC. Model 1 (including CRE, INR, WBC count and "numbers of complications") and model 3 (including MELD score and "numbers of complications") have good predictive values.
References
1 Gao B, Bataller R. Alcoholic liver disease: pathogenesis and new therapeutic targets. Gastroenterology 2011;141:1572-1585. PMID: 21920463
2 Jepsen P, Ott P, Andersen PK, Sørensen HT, Vilstrup H. Clinical course of alcoholic liver cirrhosis: a Danish population-based cohort study. Hepatology 2010;51:1675-1682. PMID: 20186844
3 Nyblom H, Berggren U, Balldin J, Olsson R. High AST/ALT ratio may indicate advanced alcoholic liver disease rather than heavy drinking. Alcohol Alcohol 2004;39:336-339. PMID: 15208167
4 European Association for the Study of the Liver. EASL clinical practice guidelines on the management of ascites, spontaneous bacterial peritonitis, and hepatorenal syndrome in cirrhosis. J Hepatol 2010;53:397-417. PMID: 20633946
5 Abdel-Khalek EE, El-Fakhry A, Helaly M, Hamed M, Elbaz O. Systemic inflammatory response syndrome in patients with liver cirrhosis. Arab J Gastroenterol 2011;12:173-177. PMID: 22305496
6 Bleibel W, Al-Osaimi AM. Hepatic encephalopathy. Saudi J Gastroenterol 2012;18:301-309. PMID: 23006457
7 Kim JN, Sohn KM, Kim MY, Suk KT, Jeong SW, Jung HE, et al. Relationship between the hepatic venous pressure gradient and first variceal hemorrhage in patients with cirrhosis: a multicenter retrospective study in Korea. Clin Mol Hepatol 2012;18:391-396. PMID: 23323255
8 Altman C, Grangé JD, Amiot X, Pelletier G, Lacaine F, Bodin F, et al. Survival after a first episode of spontaneous bacterial peritonitis. Prognosis of potential candidates for orthotopic liver transplantation. J Gastroenterol Hepatol 1995;10:47-50. PMID: 7620107
9 Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33:464-470. PMID: 11172350
10 Maddrey WC, Boitnott JK, Bedine MS, Weber FL Jr, Mezey E, White RI Jr. Corticosteroid therapy of alcoholic hepatitis. Gastroenterology 1978;75:193-199. PMID: 352788
11 Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, ter Borg PC. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology 2000;31:864-871. PMID: 10733541
12 European Association for the Study of Liver. EASL clinical practical guidelines: management of alcoholic liver disease. J Hepatol 2012;57:399-420. PMID: 22633836
13 Arroyo V, Ginès P, Gerbes AL, Dudley FJ, Gentilini P, Laffi G, et al. Definition and diagnostic criteria of refractory ascites and hepatorenal syndrome in cirrhosis. International Ascites Club. Hepatology 1996;23:164-176. PMID: 8550036
14 Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973;60:646-649. PMID: 4541913
15 Mandayam S, Jamal MM, Morgan TR. Epidemiology of alcoholic liver disease. Semin Liver Dis 2004;24:217-232. PMID: 15349801
16 Rehm J, Taylor B, Mohapatra S, Irving H, Baliunas D, Patra J, et al. Alcohol as a risk factor for liver cirrhosis: a systematic review and meta-analysis. Drug Alcohol Rev 2010;29:437-445. PMID: 20636661
17 Becker U, Deis A, Sørensen TI, Grønbaek M, Borch-Johnsen K, Müller CF, et al. Prediction of risk of liver disease by alcohol intake, sex, and age: a prospective population study. Hepatology 1996;23:1025-1029. PMID: 8621128
18 Nalpas B, Vassault A, Charpin S, Lacour B, Berthelot P. Serum mitochondrial aspartate aminotransferase as a marker of chronic alcoholism: diagnostic value and interpretation in a liver unit. Hepatology 1986;6:608-614. PMID: 3732996
19 Cooper GS, Bellamy P, Dawson NV, Desbiens N, Fulkerson WJ Jr, Goldman L, et al. A prognostic model for patients with end-stage liver disease. Gastroenterology 1997;113:1278-1288. PMID: 9322523
20 Conn HO. A peek at the Child-Turcotte classification. Hepatology 1981;1:673-676. PMID: 7308998
21 Altman DG, Royston P. What do we mean by validating a prognostic model? Stat Med 2000;19:453-473. PMID: 10694730
22 Angermayr B, Cejna M, Karnel F, Gschwantler M, Koenig F, Pidlich J, et al. Child-Pugh versus MELD score in predicting survival in patients undergoing transjugular intrahepatic portosystemic shunt. Gut 2003;52:879-885. PMID: 12740346
23 Kim WR, Biggins SW, Kremers WK, Wiesner RH, Kamath PS, Benson JT, et al. Hyponatremia and mortality among patients on the liver-transplant waiting list. N Engl J Med 2008;359:1018-1026. PMID: 18768945
24 Everhart JE, Lombardero M, Detre KM, Zetterman RK, Wiesner RH, Lake JR, et al. Increased waiting time for liver transplantation results in higher mortality. Transplantation 1997;64:1300-1306. PMID: 9371672
Received February 24, 2013
Accepted after revision May 6, 2013 |