The histological pattern of fibrosis in liver cirrhosis varies in different chronic liver diseases, and hepatic decompensation may be differentiated in consequences of fibrosis (i.e. ascites, variceal bleeding) or in lack of function (i.e. jaundice) resulting in aetiology-specific variable morbidity and mortality.
To evaluate patterns of hepatic decompensation in relation to the aetiology of liver cirrhosis.
Two different cohorts were retrospectively evaluated between 2002 and 2007. Cohort A was for hypothesis generation and consisted of 220 cirrhotic patients. To confirm the initial observations a second cohort B (n = 217) was analysed. The different patterns of hepatic decompensation evaluated were ascites, jaundice, encephalopathy, variceal bleeding, spontaneous bacterial peritonitis, hepatorenal syndrome or hepatocellular carcinoma.
Furthermore, we analysed survival in relation to pattern of decompensation in alcoholic vs. non-alcoholic liver disease.
Alcoholics were more frequently hospitalised for ascites (cohort A: 81.4% vs. 65.4%, P = 0.016; cohort B 71.3% vs. 58.5%, P = 0.085). In contrast, non-alcoholics presented with higher rates of hepatocellular carcinoma (cohort A: 23.1% vs. 11.9%, P = 0.046; cohort B 38.6% vs. 22.5%, P = 0.018). There were no significant differences in jaundice, variceal bleeding, hepatorenal syndrome or encephalopathy.
Survival was significantly impaired in non-alcoholic cirrhosis once ascites occurred (P = 0.003), whereas ascites did not predict higher mortality in patients with alcoholic cirrhosis.
Ascites is the leading initial pattern of decompensation in alcoholic cirrhosis whereas hepatocellular carcinoma dominates in non-alcoholics. Non-alcoholics developing ascites show a poor survival.
The histological pattern of fibrosis in liver cirrhosis varies in different chronic liver diseases: perisinusoidal fibrosis in alcoholic or non-alcoholic fatty liver disease vs. predominantly periportal fibrosis in chronic viral hepatitis or autoimmune liver diseases. Decompensation of liver cirrhosis can be either attributed to portal hypertension (i.e. variceal bleeding, ascites), loss of hepatic function (i.e. jaundice, hepatic encephalopathy), hepatocellular carcinoma or impaired immunity leading to bacterial infections and sepsis.[1-7] Correlation between histopathological and clinical data indicates that a histologically proven small nodule size and thick fibrous septa are associated with clinically relevant portal hypertension.
We hypothesised that the different patterns of fibrosis will lead to different patterns of decompensation. Consequently, hepatic decompensation may be dependent on the underlying aetiology of chronic liver disease.
Although management and prognosis have been separately described for leading causes of liver cirrhosis like alcoholic liver disease, hepatitis B and C,[10, 11] non-alcoholic steatohepatitis[12, 13] or autoimmune liver disease,[14, 15] the clinical course and pattern of decompensation of alcoholic vs. non-alcoholic liver cirrhosis has not been well described yet. Thus, we assessed different patterns of hepatic decompensation and their prognostic relevance in alcoholic vs. non-alcoholic liver cirrhosis and analysed cirrhotic patients at their first admission to two tertiary referral centres.
Materials and Methods
Patients with liver cirrhosis admitted to two tertiary referral centres (University of Leipzig/Germany, Hôtel Dieu Hospital, Lyon, France) between the years 2002–2007 were retrospectively analysed for patterns of first decompensation. Individuals with prior hospitalisation or malignancy other than hepatocellular carcinoma were excluded from additional investigations. Aetiology of cirrhosis was classified as alcohol-related (year-long alcohol abuse >60 g/day in men and >40 g/day in women) and non-alcohol-related. The non-alcoholic cirrhosis patients comprised: chronic hepatitis B (HBsAg positive) or hepatitis C (HCV-RNA positive), primary biliary cirrhosis (AMA-M2 positive), primary sclerosing cholangitis (pANCA positive, typical result in endoscopic retrograde cholangiography), autoimmune hepatitis (according to the international autoimmune hepatitis score), non-alcoholic fatty liver disease (steatosis in ultrasound, exclusion of the other mentioned aetiologies), Wilson′s disease (copper excretion in 24 h urine, histology or Kayser–Fleischer ring), hemochromatosis (transferrin saturation >45%, elevated ferritin, HFE mutation), cryptogenic cirrhosis or cirrhosis of other aetiology.
Hepatic decompensation has been defined as occurrence of one of several complications of liver cirrhosis according to the Child-Pugh-score. We investigated the following patterns of decompensation: ascites (confirmed by ultrasound), jaundice (clinical evaluation or bilirubin >3 times the upper limit of normal), variceal bleeding (endoscopic report), encephalopathy (clinical diagnosis, number connection test) or hepatocellular carcinoma (biopsy proven or typical contrast medium performance in two radiographic evaluations), respectively. Multiple clinical events could be observed in single cases at the first episode and were separately investigated during additional analysis.
Hepatorenal syndrome (verified by treatment with albumin and terlipressin), spontaneous bacterial peritonitis (leucocytes >500/μL ascites in combination with antibiotic and albumin therapy), bacterial infections (elevated leucocytes and C-reactive protein in combination with antibiotic treatment) or pneumonia (radiologic report in combination with antibiotic therapy) were recorded as possible additional complications.
A total of 220 cirrhotic German patients from one ward were evaluated as cohort A. Results were confirmed in a second cohort of German patients from another ward from the same hospital and French patients (cohort B, n = 217). This stepwise approach was chosen to generate hypotheses in cohort A and to subsequently confirm the results of cohort A in the independent cohort B and in the overall analysis.
Statistical analyses were carried out using pasw Statistics 18 (version 18.0.1; PASW Inc., Chicago, IL, USA). Fisher′s exact and chi square test, and Mann–Whitney U-test in case of non normally distributed variables were used where appropriate. For the hypothesis generating evaluation of cohort A a P-value of 0.05 was considered statistically significant, while we accepted a slightly higher P-value of P < 0.1 for the confirmatory cohort B as significant. Patient′s survival was calculated with Kaplan–Meier curves and was only performed in German patients.
Cox regression survival analysis was performed including age, gender, HCC, ascites, Child status and aetiology of liver disease in a univariate and multivariate analysis, as well as a similar analysis in which aetiology was not included as a parameter but alcoholic and non-alcoholic liver disease were explored independently. The chosen level of statistical significance was P < 0.05.
Importantly, alcohol consumption after discharge from hospital could not be included in the mathematical model because data on drinking habits were not available in the follow-up period.
The study was approved by the local ethics committee of the University of Leipzig (registration number 098-2008).
Cohort A consisted of 220 German patients, while cohort B included 217 individuals from Germany (n = 134) and France (n = 83) (Table 1). Overall, 70.3% of patients were men and suffered from Child B or C cirrhosis in 30.2% and 44.6% of cases, respectively.
Table 1. Baseline characteristics of the study cohorts
Cohort A (n = 220)
Cohort B (n = 217)
56.2 ± 11.2
58.5 ± 13.1
Duration of hospitalisation (days)
13.54 ± 8.8
13.1 ± 10.1
Alcoholic cirrhosis (n = 328)
Male (n = 249), female (n = 79)
Chronic hepatitis C (n = 30)
Male (n = 21), female (n = 9)
Chronic hepatitis B (n = 11)
Male (n = 10), female (n = 1)
Cryptogenic cirrhosis (n = 37)
Male (n = 16), female (n = 21)
Primary biliary cirrhosis (n = 7)
Male (n = 0), female (n = 7)
Primary sclerosing cholangitis (n = 3)
Male (n = 2), female (n = 1)
Autoimmune hepatitis (n = 5)
Male (n = 1), female (n = 4)
Non-alcoholic steatohepatitis (n = 5)
Male (n = 3), female (n = 2)
Wilsonˋs disease (n = 2)
Male (n = 0), female (n = 2)
Hemochromatosis (n = 2)
Male (n = 2), female (n = 0)
Other aetiologies (n = 7) (i.e. drug-induced, toxic)
Male (n = 3), female (n = 4)
Alcoholic liver cirrhosis (75.1%) was the leading aetiology in both cohorts. Viral hepatitis occurred significantly more frequently in cohort B, because 70% of chronic hepatitis C cases were recruited in the French population (P < 0.001).
Patients with alcoholic liver cirrhosis were significantly more often men (cohort A: 75.9% (n = 126/168) vs. 46.2% (n = 24/52), P < 0.001; cohort B: 76.9% (n = 123/160) vs. 59.6% (n = 34/57), P = 0.013), younger (cohort A: 55.2 ± 10.3 vs. 59.5 ± 13.2 years, P = 0.014; cohort B: 56.1 ± 12.7 vs. 65 ± 12 years, P < 0.001), and smokers (cohort A: 58.1% (n = 79/136) vs. 29.5% (n = 13/44), P = 0.001; cohort B: 64.4% (n = 65/101) vs. 27.3% (n = 6/22), P = 0.001) compared with subjects with non-alcoholic disease.
Pattern of hepatic decompensation in alcoholic vs. non-alcoholic cirrhosis
Ultrasound reports were available in 95.2% of cases. Examined by ultrasound, patients with alcoholic cirrhosis developed ascites significantly more frequently than cases with non-alcoholic aetiology (Table 2). If ascites was quantified in ultrasound examination, severe ascites occurred significantly more often in alcoholics (Figure 1).
Table 2. Pattern of decompensation in alcoholic vs. non-alcoholic cirrhosis
Pattern of decompensation in alcoholic vs. non-alcoholic cirrhosis
P = 0.016
P = 0.085
Spontaneous bacterial peritonitis in patients with ascites
P = 0.261
P = 0.348
P = 0.046
P = 0.018
P = 0.785
P = 0.426
P = 0.039
P = 0.085
P = 0.015
P = 0.84
P = 0.752
P = 0.343
Subsequent spontaneous bacterial peritonitis in individuals with ascites tended to be more frequent in alcoholic vs. non-alcoholic liver disease, although differences were not significantly different (20% vs. 9.5% (P = 0.261) in cohort A; 11.8% vs. 4.8% (P = 0.689) in cohort B) (Table 2). In contrast, hepatocellular carcinoma was the dominating pattern of decompensation in non-alcoholic cirrhosis (P = 0.046 and P = 0.018 respectively) (Table 2).
The prevalence of jaundice, hepatic encephalopathy and variceal bleeding did overall not differ significantly between alcoholic and non-alcoholic cases. Only in cohort A, jaundice occurred more frequently in non-alcoholic cirrhosis and hepatic encephalopathy in alcoholic disease; however, these results could not be confirmed in cohort B (Table 2).
Infectious complications occurred in 19.7% (n = 86/437) of cases. Urinary tract infections (38.8%, n = 34/86), spontaneous bacterial peritonitis (37.2%, n = 29/86), pneumonia (15.1%, n = 13/86) and sepsis (13.8%, n = 11/86) were most frequently observed. Patients with alcoholic liver cirrhosis tended to develop infections more frequently than individuals with non-alcoholic diseases. This tendency was mainly caused by the prevalence of spontaneous bacterial peritonitis (Tables 2 and 3).
Table 3. Infectious complications in alcoholic vs. non-alcoholic cirrhosis
Infectious complications in alcoholic vs. non-alcoholic cirrhosis
P = 0.068
P = 0.546
Urinary tract infection
P = 0.044
P = 0.79
P = 0.056
P = 1.0
P = 0.36
P = 1.0
Prognosis of hepatic decompensation
Duration of hospitalisation (cohort A 13.4 ± 8.6 vs. 13.9 ± 9.4, P = 0.911; cohort B 13.2 ± 9.4 vs. 12.8 ± 11.9, P = 0.181) and in-hospital mortality (cohort A: 6.5% vs. 1.9%, P = 0.301; cohort B: 6.4% vs. 9.4%, P = 0.538) were not different between alcoholic and non-alcoholic cirrhosis.
Due to missing information in the electronic French documentation, post hospital survival could only be assessed in German patients (n = 354; alcoholic cirrhosis: n = 278; non-alcoholic disease: n = 76). Child B cases survived longer than individuals with Child C cirrhosis (1639 vs. 1198 days; P = 0.006). A significant difference between alcoholics and non-alcoholics was not observed within Child B or C individuals respectively. Mean survival was 1441 ± 93.4 days in alcoholics and 1264 ± 168.8 days in non-alcoholics (P = 0.404).
If survival was analysed according to the pattern of hepatic decompensation, only ascites showed a significant prognostic relevance: Individuals with alcoholic cirrhosis and ascites survived longer than cases with non-alcoholic liver diseases and ascites (median 668 vs. 198 days; P = 0.031) (Figure 2a).
Within the two aetiologies, ascites limited survival only in non-alcoholics: They experienced a poor survival in case of hydropic decompensation compared with cases without ascites (198 vs. 2572 days; P = 0.003). In contrast, alcoholics with ascites did not significantly die earlier (668 vs. 1459 days; P = 0.251) (Figure 2b,c).
In a Cox regression analysis, survival in alcoholic cirrhosis was impaired in univariate analysis by age, Child status, and hepatocellular carcinoma, but not by ascites. In contrast, patients with non-alcoholic disease died earlier in case of ascitic decompensation (Table 4).
Table 4. Parameters influencing survival in cirrhosis: univariate and multivariate analysis
Multivariate analysis confirmed the observations for age, Child status and hepatocellular carcinoma in alcoholic cirrhosis, and for ascites in non-alcoholic liver cirrhosis (Table 4).
The present study analysed differences between the first decompensation in alcoholic vs. non-alcoholic liver cirrhosis, because the clinical management and prognosis of cirrhotic patients may differ according to aetiology and subsequent complications after hepatic decompensation.
Patients with alcoholic liver cirrhosis predominated in both investigated cohorts and were significantly younger than cases with non-alcoholic disease. These different baseline characteristics can be explained by several epidemiological and medical observations. Alcohol is the leading aetiology in several studies investigating the course of liver cirrhosis.[18, 19] In Saxony, up to 23% of the general population consume alcohol at a harmful level. In contrast, the prevalence of chronic viral hepatitis B and C in Eastern parts of Germany is low, and significant lower than in matched control groups from Western Germany.[21, 22]
Medical reasons for the differences between baseline characteristics are the slow progression during the natural course of viral liver diseases[11, 23-25] and effective antiviral treatment options which prevent development and decompensation of cirrhosis.[26-31]
Individuals with alcoholic liver cirrhosis developed significantly more often and more severely ascites than cases with non-alcoholic liver diseases. These results corroborate data from a recent population based Danish and a monocentric Norwegian study which observed ascites as the leading initial hepatic decompensation pattern in 55% and 67% of cases respectively.[9, 32] If alcoholic and hepatitis C-related cirrhosis are directly compared with each other, a Spanish trial likewise observed higher rates of ascites in alcoholic cirrhosis (59% vs. 48%). Patients with viral hepatitis still can have dangerous drinking habits and may in fact have alcoholic cirrhosis and viral hepatitis,[34, 35] however, none of the non-alcoholic liver disease patients in our cohorts reported dangerous drinking habits. Differences in the nutritional status did not appear to be related to the higher frequency of ascites in alcoholic cirrhosis, because we did not observe differences in serum albumin levels in alcoholic (30 g/L) vs. non-alcoholic liver disease (31.5 g/L).
In contrast to the decompensation pattern in alcoholic cirrhosis, hepatocellular carcinoma was the dominating complication in non-alcoholic cirrhosis. These data are corroborated by a large recent population based analysis which observed non-alcoholic steatohepatitis, diabetes, and hepatitis C as leading underlying aetiologies before alcoholic cirrhosis in patients with hepatocellular carcinoma. If hepatitis C-related cirrhosis is compared with cirrhosis as a consequence of alcohol abuse or non-alcoholic steatohepatitis, the oncogenic risk of the virus is further elucidated.[33, 37, 38]
Although some authors did not observe a difference in 10 years survival between alcoholic and non-alcoholic liver cirrhosis, other investigators described a better or poorer survival for alcoholics compared with other aetiologies. These differences may be related to the already discussed different treatment options in non-alcoholic liver diseases and various rates of abstinence in different cohorts, however, recent data indicate that the initial pattern of decompensation is critical for further progression of alcoholic liver cirrhosis. The 5-year death rate in alcoholic cirrhotics without complications, with ascites, with variceal bleeding, with the combination of ascites plus variceal bleeding, and with encephalopathy is 58%, 59%, 64%, 80% and 85% respectively. In our trial, ascites was a significant disadvantage in non-alcoholic disease (2572 vs. 198 days; P = 0.003). These data deliver the clinical correlate to the British mathematical model which observed a transition from stage 1 cirrhosis (no ascites, no varices) to stage 3 (cirrhosis with ascites ± varices) more frequently in alcoholic vs. non-alcoholic cirrhosis, while cases with non-alcoholic cirrhosis died faster than alcoholics. Furthermore, patients with decompensated cirrhosis and untreated hepatitis B infection die in 30% of cases after ascitic decompensation, but only in 8% after variceal bleeding. Thus, occurrence of ascites in non-alcoholic cirrhosis should be considered as a negative prognostic indicator and should lead to timely evaluation for liver transplantation.
Although we could detect differences in the pattern of hepatic decompensation between alcoholic and non-alcoholic liver cirrhosis, the data deserve cautious interpretation and additional confirmation in standardised prospective cohorts: Due to the retrospective study design, some complications were not always uniformly documented in patient′s charts (i.e. hepatorenal syndrome) or subject of the individual estimation of the treating physician (i.e. encephalopathy, jaundice). Follow-up data on survival were only available for German patients due to missing information in the French database. Data on sustained or reduced alcohol consumption after discharge from hospital could not be assessed. Nevertheless, the presented data indicate prognostic parameters on patient′s survival after presentation to two tertiary referral centres: Ascites was the leading pattern of hepatic decompensation in alcoholic cirrhosis, while hepatocellular carcinoma dominated in non-alcoholic diseases. If patients with non-alcoholic cirrhosis developed ascites, they showed significantly impaired survival.
Declaration of personal and funding interests: None.