Survival of Patients with Alcohol-Related Liver Disease Cirrhosis - Usefulness of the New LIV-IN Prognostic Score

1. Introduction

Alcohol consumption is associated with the occurrence of a large number of diseases such as alcohol-related liver disease (ALD), arterial hypertension, cardiomyopathy, acute and chronic pancreatitis, gastritis, esophageal cancer, etc [1]. According to the World Health Organisation (WHO) data, excessive alcohol consumption is the main cause of death of 3 million people in 2016 year [2]. Alcohol may lead to direct toxic liver damage, and cause liver steatosis, steatohepatitis, cirrhosis, and hepatocellular carcinoma (HCC). In Europe, the mortality rate due to ALD cirrhosis is 9.2/100 000 inhabitants [3]. In 2019, out of the total number of deaths due to liver disease, one-fourth was related to ALD cirrhosis [4]. Patients with decompensated liver cirrhosis have a 9.7 higher risk of death compared to the general population [5]. Infection, especially spontaneous bacterial peritonitis (SBP) and sepsis, as well as hepatorenal syndrome (HRS), increase the risk for mortality 10-20 times. Hepatic encephalopathy (HE), variceal bleeding, and ascites are also associated with decreased survival [3]. However, it has been proven that active alcohol consumption is the main predictor of survival in patients with ALD cirrhosis because it may lead to increased bacterial translocation and liver inflammation with consequently higher intrahepatic resistance and multiorgan failure [6].

In daily clinical practice, the most frequently used prognostic scores are the Model for end-stage liver disease (MELD), the Model for end-stage liver disease-Sodium (MELD-Na), and the Child Pugh score [7,8]. Since 2013, the chronic liver failure-sequential organ failure assessment (CLIF-C SOFA) score for critically ill patients has been in use, as well as the Chronic Liver Failure Consortium-acute-on-chronic liver failure (CLIF-C ACLF) score for patients with ACLF, and Chronic Liver Failure Consortium-C acute decompensation (CLIF-C AD) for patients with acute decompensation without ACLF [9].

The well-known feature of liver cirrhosis is chronic systemic inflammation, mainly related to bacterial translocation [10]. The neutrophil-to-lymphocyte ratio (NLR) is a simple, well-known marker of systemic inflammation with previously documented usefulness as a prediction tool in patients with liver cirrhosis, both acute decompensation (AD) or acute-on-chronic liver failure (ACLF) [11].

Another novel prognostic score is albumin-bilirubin (ALBI), which can reflect early liver function deterioration. Firstly, it was used for patients with HCC, and later it was proven as a good survival predictor for other non-malignant chronic liver disease [12].

Patients diagnosed with decompensated liver cirrhosis are more prone to bacterial infection due to immune dysfunction, increased gut permeability, intestinal bacterial overgrowth, gut dysbiosis, etc. According to previously published studies, approximately 32-40% of inpatients had community-acquired (CA) or healthcare-associated infections (HAIs), which further decreased short-term survival [13].

Based on this consideration, the present study aimed to investigate 28-day mortality in patients with ALD cirrhosis, as well as identify factors associated with decreased short-term survival. Moreover, this study investigated the diagnostic accuracy of the newly created original LIV-IN (LIVer mortality INpatients) score and the following „old“ prognostic scores MELD, MELD-Na, ALBI, NLR, as well as CLIF-C AD in patients with acute decompensation, and CLIF C-ACLF in patients diagnosed with ACLF.

2. Materials and Methods

2.1. Study Design and Inclusion Criteria

A prospective cohort observational study was conducted at the Gastroenterology and Hepatology Intensive Care Unit of the Emergency Center and the Semi-intensive Care Unit of the Clinic for Gastroenterohepatology, University Clinical Center of Serbia. The study included hospitalized subjects with ALD cirrhosis, who were treated in our institution from November 2022 to August 2023.

The diagnosis of ALD cirrhosis was made based on anamnestic and/or hetero-anamnestic data on long-term harmful alcohol consumption, as well as based on typical laboratory analyses, clinical signs, radiological features, and after excluding other viral, immunological, and metabolic causes of liver disease. The exclusion criteria were as follows: age <18 years, HCC, acute or chronic decompensation of another extrahepatic system that may be related to multiorgan failure, extrahepatic cancers, severe trauma, human immunodeficiency virus (HIV) infection, and pregnancy. All patients were followed for 28 days. The study was approved by the Ethics Committee of the University Clinical Center of Serbia and was conducted according to the principles of the Helsinki Declaration (protocol code: 17/4; date of approval: 26.01.2023.).

2.2. Data Collection

The following data were from the patient’s medical records: age, gender, presence of ascites, HRS, HE, upper gastrointestinal bleeding (UGIB), and systemic inflammatory response syndrome (SIRS), as well as body temperature, respiratory rate, oxygen saturation (SpO2), arterial pressure, and data regarding active alcohol consumption. In each patient, the initial diagnostic work-up included a chest X-ray, abdominal ultrasound, complete blood count, comprehensive metabolic panel, and microbial cultures (venous blood and urine). In cases in which diagnostic paracentesis could be performed, peritoneal fluid was sampled and sent for biochemical, cytological, and microbial analyses.

According to the National guidelines for diagnosis and treatment of alcohol use disorder (AUD), issued by Serbian Ministry of Health, active and excessive alcohol consumption was defined as alcohol intake >14 standard drinks weekly, or >4 standard drinks on one occassion in men, and >7 standard drinks weekly, or >2 standard drinks on one occassion in women, at least during the last three monts. One standard drinks is defined as containing 13 grams of alcohol (one small bottle of beer (330 ml, 5% of ethanol), one glass of wine (120 ml, 12% of ethanol), and one glass of spirits (40 ml, 40% of ethanol)) [14].

According to the European Association for the Study of the Liver (EASL) criteria, ascites was graded as: „mild“ or grade I ascites - if it was detectable only by ultrasound; „moderate“ or grade II - if the clinical examination showed symmetrical abdominal distension with positive clinical signs for ascites; and „large“ or grade III – a large amount of free fluid with visible abdominal distension (15). EASL criteria were also applied for the diagnosis of HRS and SBP [15].

The West Heaven criteria were used for grading HE. Patients with minimal HE and grade I HE according to the West Haven criteria were classified as covert HE, while patients with HE grades II, III, and IV according to the West Haven criteria were classified as overt HE. In patients who did not meet the West Heaven criteria for grade II, the animal naming test was used to establish the diagnosis of covert encephalopathy. In the animal naming test, the patients were asked to name as many animals as possible within 1 minute. If the subjects named less than 10 animals in the specified period, the diagnosis of covert hepatic encephalopathy was made [16].

Patients diagnosed with infection on admission or within the first 48 hours of admission were considered to have a CAI. If the infection was proven by diagnostic processing after 48 hours, the patient was considered to have a HAI.

The following criteria were used to diagnose a urinary tract infection (UTI): abnormal urine sediment (>10 WBC per microscopic field of view) with a positive urine culture (>100.000 Colony Forming Unit (CFU)). Based on the clinical findings and chest X-ray observed inflammatory infiltrates in the lung, pneumonia was diagnosed. The term spontaneous bacteremia was defined as the presence of a positive blood culture without an evident primary focus of infection.

All patients who were hospitalized due to UGIB underwent upper endoscopy. Based on the upper endoscopy findings, bleeding was categorized as variceal, ulcer, or other, in cases when other pathological changes were identified in the upper portion of GI such as portal hypertensive gastropathy, Mallory-Weis syndrome, gastric antral venectasia, Dilafou lesion, etc.

The patients were diagnosed with SIRS if they met two of the following four criteria:1. body temperature >38^oC or <36 ^oC, 2. heart rate >90/minute, 3. respiratory rate >20/minute, and WBC >12x10⁹ or <4x10⁹.

The term AD of liver cirrhosis was considered in case of complications of the underlying disease such as hepatic encephalopathy, ascites, variceal bleeding, bacterial infection, or as a combination of the aforementioned complication.

The term ACLF was used in cases of severe acute decompensation of liver cirrhosis with functional failure of one or more of the following six organ systems: liver, kidney, brain, coagulation, circulation, and respiratory system. To stratify the subjects, the European Association for the Study of the Liver (EASL)-Chronic Liver Failure (CLIF) Consortium (CLIF-C) ACLF criteria were used [17].

According to the previously recorded data following prognostic scores were calculated: MELD, MELD-Na, NLR, ALBI. In patients with acute decompensation without ACLF, the CLIF-C AD score was calculated, as well as the CLIF-C ACLF score in patients with ACLF diagnosis.

In the ACLF group, if one organ failure system was identified, the patients were further classified into grade 1, for failure of two organ systems into grade 2, and three or more organ failures into grade 3.

2.3. Follow-Up

All patients were followed up for 28 days. Subjects lost in the follow-up for different reasons were excluded from the present study. Also, if a hepatic or extrahepatic tumor diagnosis was made during the mentioned follow-up period, the patients were excluded from the study. In case of a fatal outcome, the cause of death was recorded.

2.4. Statistical Analysis

Numerical variables with normal distribution were expressed as mean ± standard deviation (SD), as well as median with a 25-75% interquartile range in cases where the criteria for normal distribution were not fulfilled. Nominal and ordinal variables are presented as absolute numbers and percentages. The Kolmogorov-Smirnof test was used to check the distribution normality. For the analysis of categorical variables, r x k contingency tables, Fischer’s test, and Person Chi-square test were used. Numerical data with normal distribution were analyzed using the Student’s T-test and ANOVA, and if the distribution was not normal, the Mann-Whitney test was used. The ROC curve was used to analyze the clinical accuracy of prognostic scores. The identification of risk factors for the occurrence of a lethal outcome was carried out using logistic regression. Statistical significance is defined as p<0.05. Statistical analysis was performed using the statistical software IBM SPSS, version 20.

3. Results

After the initial diagnostic work-up, a total of 150 patients who met the inclusion criteria were enrolled in the study. During the following period, 5 patients were excluded due to different reasons, and the final number of all included patients in our study was 145. Based on the CLIF-C-OF score, the patients were further divided into two groups: acute decompensation (AD) (n=107) and ACLF (n=38) (Figure 1).

Baseline demographic and clinical characteristics are presented in Table 1. Among the patients included in this study, 129 (89%) were male, with a mean age of 56.42±10.87 years. No difference in age was recorded between patients with AD and ACLF, while the frequency of the female gender was significantly higher in ACLF patients, compared to the AD group (p=0.03). On admission, 123 (84.8%) patients were actively drinking alcohol. Eighty-three (57.2%) patients were admitted to the ICU. ICU admission was more common in ACLF, compared to AD patients (73.7% vs. 51.4%, p=0.02)

The vast majority of patients (80.7%) had ascites. Patients with ACLF had ascites more commonly and also had ascites of higher grade compared to patients with AD (p=0.009, p<0.001, respectively).

Nighty seven patients (66.9%) were diagnosed with HE, among them 87.6% (n=85) of patients had overt, and 12 (12.4%) had covert hepatic encephalopathy. Hepatic encephalopathy was more commonly detected in ACLF patients compared to those with AD (86.8% vs. 59.8%, p=0.02). Overt form of hepatic encephalopathy was more frequently reported in AD patients (90.6% vs 81.8%, p=0.04)

Upper gastrointestinal bleeding was seen in 55(37.9%) cases. Thirty-one patients (56.4%) were diagnosed with variceal bleeding, 4 (7.3%) with ulcus bleeding, and other causes of UGIB were seen in 20 (36.4%) of patients. There were no differences in terms of UGIB between the compared groups.

HRS were diagnosed in 19 patients, among them 9 (8.5%) were in AD, and 10 (26.3%) were in the ACLF group (p=0.01). SIRS was seen in 38(26.2%) patients, and it was more frequent in the ACLF group ( 39.47% vs 21.49%, p=0.03).

Baseline laboratory analyses are presented in detail in Table 2. In the ACLF group, the following laboratory analyses were significantly higher than in the AD group: WBC, total bilirubin, creatinine, CRP, fibrinogen, procalcitonin, PT, and INR (p<0.01). Serum albumin level was statistically lower in the ACLF group (p<0.05).

Forty-five patients (31.03%) were diagnosed with CAIs. There were no statistical differences in the frequency of CAIs among AD and ACLF groups (n=32 (29.9%); n=13 (34,2%), respectively, p=0.07). The most common CAI was UTI (n=21; 46.7%) followed by pneumonia (n=14; 31,1%). Four patients (8.9%) were diagnosed with pneumonia and UTI, and bacteremia was seen in another four cases (8.9%). The most frequent CAI in the AD group was UTI (n=21; 46,7%), while pneumonia was the most common in the ACLF group ( n=6; 46.2%) (Table 3).

Health-care-associated infections were found in 63 patients (43.4%), among them 41 (38.3%) were from the AD group and 22 (57.9%) were from the ACLF group (p=0.05). As shown in Table 3, the following types of infections were diagnosed: UTI (n=35; 55.6%), bacteremia (n=20; 31.7%), pneumonia (n=5; 7.9%), SBP (n=2; 3.2%) and combined UTI and pneumonia (n=1; 1.6%). Urinary tract infection was the most frequent primary infection focus in the AD group ( n=25; 61%). Bacteremia, as well as UTI, were the most frequently observed HAI in the ACLF group.

3.1. Twenty-Eight Days Survival of Patients with ALD Cirrhosis

During the 28-day follow-up, the fatal outcome occurred in 46 (31.7%) patients. The mortality rate was significantly higher in the ACLF group (n=22, 57.9%) compared to the AD group (n=24, 22.4%) (p<0.01). In the ACLF group, the mortality rate in terms of ACLF gradus was as follows: gradus I: n=6 (42.9%); gradus II: n=8(57.1%); gradus III: n=8 (88.9%).

The main causes of death are presented in Table 4. Liver-related death occurred in 15 (32.5%) patients while the rest was due to another extrahepatic reason. The most frequently observed cause of death was heart failure (n=17; 37%), followed by lung failure (n=13; 28.3%). Liver-related deaths due to variceal bleeding, infections, and other liver-related complications in the AD group were seen in 9 (37.5%) patients, while this number was lower in the ACLF group (n=6, 27.2%).

The main predictive factors associated with 28-day mortality in univariate analysis were: HE (OR 4.912, p=0.01), CAIs (OR 2.3, p=0.023), ascites (OR 7.836, p=0.007), HRS (OR 6.040, p=0.001), SIRS (OR 8.775, p<0.01), and ACLF ( OR 7.25, p<0.01). When we further enrolled all these statistically significant variables from univariate analyses into multivariate analyses, we found that this model is statistically significant (χ 2 59.361, df =7, p<0.01). From a total of six variables, only the following four remained statistically significant: HE (OR 3.63, p=0.04), ascites (OR 6.896, p=0.03), SIRS (OR 9.323, p<0.01), and ACLF (OR 3.539, p=0.013) (Table 5).

3.2. Prediction of 28-Day Mortality Based on Prognostic Scores-Formulation of the New LIV-IN Prognostic Score

According to the previously mentioned predictive factors, a new original prognostic score named LIV-IN (LIVer mortality INpatients) for inpatients with decompensated ALD cirrhosis was created. In this predictive model, we enrolled the following variables: CAIs, HE, ascites, HRS, SIRS, and fibrinogen. Variables were scored as follows:

1. Community-acquired infection: Yes (1 point); No (0 point);

2. Hepatic encephalopathy: Yes (1 point); No (0 point);

3. Ascites; without (0 point); mild (1 point); moderate (2 points); large (3 points);

4. HRS: Yes (1 point); No (0 point);

5. SIRS: Yes ( 1 point); No (0-point);

6. fibrinogen (g/l).

The score was calculated as a sum of all points with serum fibrinogen value.

The present study examined the clinical accuracy of the LIV-IN predictive model and previously known following scores in predicting 28-day mortality: MELD, MELD-Na, ALBI, NLR. For patients with AD without ACLF, we also evaluated the CLIF-C AD score, as well as the CLIF-C ACLF score for patients with ACLF.

Patients with ACLF had statistically higher MELD, and MELD-Na scores as well as LIV-IN score (p<0.01). There were no differences in NLR scores among groups. ALBI score was significantly lower in the ACLF group (Table 6).

The statistically significant discriminative power of five evaluated scores (LIV-IN, MELD, MELD-Na, NLR, and ALBI) was observed in predicting 28-day mortality in patients with ALD cirrhosis. The ROC curve is shown in Figure 2, and the summary results in Table 7. According to the results of the ROC curve, the highest predictive power was observed for the LIV-IN score (AUC 73.4%). The estimated ROC curve for MELD, MELD-Na, ALBI, and NLR scores suggests that their predictive power is 66.1%, 69.6%, 69.6%, and 61% respectively.

We further evaluated the clinical accuracy of the prognostic scores in predicting 28-day mortality in the AD group. The following scores in this study group were examined: LIV-IN, MELD, MELD-Na, ALBI, NLR, and CLIF-C AD. Statistically significant discriminative power had LIV-IN score (p<0.007, AUC 0.686, CI 0.577-0.795 ), CLIF-C AD (p=0.004, AUC 0.699, 95% CI (0.574-0.824)), ALBI (p=0.007, AUC 0.685, 95% CI (0.556-0.814)) and MELD-Na ( 0.04, AUC 0.635, 95% CI (0.500-0.770)). A ROC curve of examined scores is shown in Figure 3.

4. Discussion

The primary aim of the present study was to analyze the 28-day survival of hospitalized patients with decompensated ALD cirrhosis. During the 28-day follow-up, the lethal outcome occurred in 46 (31.7%) patients. Lethal outcome was significantly more common in the ACLF group (n=22, 57.9%) compared to the AD group (n=24, 22.4%). In the ACLF group increases in ACLF grades were followed by increases in mortality and these results are in accordance with previously published data [18,19,20]. All patients who were diagnosed with ACLF in our study had cirrhosis, which may also affect survival as shown in the study of Thanapirom K et al. [21]. Thanapirom K et al. also suggested that patients with ACLF with cirrhosis had better outcomes compared to ACLF patients without cirrhosis, explained by an inappropriate immune response which could result in less organ damage [21]. The twenty-eight-day mortality in acute decompensation of liver cirrhosis regardless of cirrhosis etiology was 20.8% in previously published data, which was similar to the results of the present study [22]. However, the results are not unique, so in the study that was conducted on patients who were hospitalized through emergency admission, it was observed that in-hospital mortality was lower and amounted to 15.9%. In addition, the average value of the MELD score in that study was also lower compared to our results [23].

Alcohol consumption, liver cirrhosis itself, as well as the use of drugs for the treatment of decompensated liver cirrhosis, can lead to heart failure and malignant rhythm disorders, and the same can lead to death [24,25]. Analyzing the causes of death in our population, it was observed that heart failure with malignant rhythm disorder is one of the most common causes of death. The present study identified other causes of death as respiratory failure due to pulmonary embolism, acute respiratory distress syndrome, pneumonia, then sepsis, bleeding with shock, etc. The leading causes of death in patients with ACLF were hemorrhagic shock, and respiratory failure as was shown in recently published data by Liu LX et al. [26]. The development of acute pancreatitis with necrosis, heart failure, heart attack, mesenteric thrombosis, and renal failure can directly lead to death in subjects with ALD cirrhosis [27]. The above data imply that a multidisciplinary approach is needed in the treatment of critically ill patients with ALD cirrhosis.

In daily clinical practice, it is very important to identify risk factors for fatal outcomes. The main predictive factors associated with 28-day mortality in univariate analysis in our study were: HE, community-acquired infections, ascites, SIRS, and ACLF. When we further enrolled all these statistically significant variables from univariate analysis into multivariate analyses, only the following four remained statistically significant: HE, ascites, SIRS, and ACLF. In the study by Trifan et al., which was conducted on more than 1000 subjects with ALD cirrhosis, it was observed that variceal bleeding, infection, SBP, sepsis, HE, ascites, HCC, and HRS were associated with an unfavorable outcome of the first hospitalization [3]. Other studies identified MELD score, lactates, infection, albumin, CRP and CRP/albumin ratio as independent predictors of in-hospital mortality [23,28]. Also, a studies examining only patients with ACLF showed that independent predictors of 28-day mortality in patients with ACLF were MELD score >26, ACLF grade 3, need for ventilation, shock and use of hemodialysis [20,29]. Based on all previous studies, as well as the results of our research, it can be concluded that hospitalized patients with ALD cirrhosis are a real challenge in treatment and follow-up.

Nowadays, mathematically prognostic scores are used in all areas of medicine in daily clinical practice for easier assessment of patients’ prognoses. In hepatology, CP and MELD are the two scores that were first introduced into clinical practice, and are still very valuable tools, even with slight modifications (eg. MELD-Na). In recent years, based on the results of the CANONIC study, two more scores have been used in clinical practice. The CLIF-C ACLF score is a clinically relevant, validated score that is superior in predicting mortality compared to the MELD and MELD-Na score for patients with ACLF, while the CLIF-C AD score is more relevant than other scores for predicting mortality in hospitalized patients with acute decompensation [30,31]. The calculation of the ALBI score uses bilirubin and albumin and therefore shows the degree of liver damage. It has been shown that it can be a reliable score for predicting survival in HCC but also for patients with ACLF [32,33].

In this study, we evaluated the discriminative power of MELD, MELD-Na, ALBI, and NLR scores in predicting 28-day mortality. Also, we created a new original predictive model named LIV-IN score for ALD cirrhosis based on the following variables: HE, ascites, SIRS, HRS, community-acquired infection, and fibrinogen. According to the results of the ROC curve, the highest predictive accuracy was observed for the new LIV-IN prognostic score (AUC 73.4%). The estimated ROC curve for MELD, MELD-Na, ALBI, and NLR scores suggests that their predictive power is 66.1%, 69.6%, 69.6%, and 61%.

Diagnostic accuracy of MELD and MELD-Na scores were lower in our study compared to the results of previously published data [34,35,36].

In the group of patients with acute decompensation, we examined all the aforementioned scores along with the CLIF-C AD score. Statistically significant discriminative power had LIV-IN score (AUC 0.686), CLIF-C AD ( AUC 0.699), ALBI (AUC 0.685), and MELD-Na (AUC 0.635).

In the ACLF group, only the LIV-IN score showed statistically significant discriminative power in predicting 28-day mortality in patients with ACLF (p=0.01, AUC 0.742), which may be due to a small number of included patients with ACLF.

Patients with AD and particularly those with ACLF have a conspicuous systemic inflammatory response and a high chance of dying [37]. However, even though SIRS was described in 42.7 % of patients with decompensation of liver cirrhosis in the previously published data, and was associated with alcohol-related liver disease, its diagnostic value in this specific group of patients remains questionable due to the influence of beta blockers, hypersplenism, and hyperventilation in liver cirrhosis [38]. In the present study, 38 (26.2%) patients were diagnosed with SIRS, and the same was more frequent in the ACLF group. The current study identified SIRS as a predictive factor of short-term mortality, and because of that, it was one of the variables that was included in the calculation of the LIV-IN score.

As mentioned earlier, one of the variables from the LIV-IN score was community-acquired infection, and this is the first predictive model that implemented bacterial infection in the predictive models. In addition to causing liver disease, alcohol consumption also increases the risk of systemic bacterial infections. It acts on T-cells in the skin and leads to infections caused by Staphylococcus aureus [39]. In addition, alcohol directly leads to mucosal damage at the level of the gastrointestinal tract and bacterial overgrowth, which leads to increased bacterial translocation. Alcohol consumption also affects the alveolar epithelium and increases the risk of pneumonia [40]. On the other hand, liver cirrhosis is characterized by immune dysfunction and excessive activation of pro-inflammatory cytokines, malnutrition, which makes these patients susceptible to infections [40,41,42]. Although the cause of acute decompensation, especially ACLF, is very often a bacterial infection, the diagnosis is not always easy due to the present impaired liver function, pronounced pro-inflammatory response, hypersplenism, abdominal distension, negative cultures, altered state of consciousness, and early recognition and treatment of the infection is crucial [43].

In our study, 45 patients (31.03%) were diagnosed with community-acquired infections, and 63 patients (43.4%) were diagnosed with healthcare-associated infections. The primary focuses of infection were as follows: UTI, pneumonia, spontaneous bacteremia, and SBP.

5. Conclusion

Hospitalized patients with ALD cirrhosis have a high 28-day mortality rate. According to the results of our study, the main cause of death was cardiovascular event, respiratory failure, and liver-related deaths. In univariate analysis, predictive factors of lethal outcome were: community-acquired infections, ascites, HE, HRS, SIRS, and ACLF. In multivariate analysis, only HE, ascites, SIRS, and ACLF were statistically significant. The new LIV-In score for the prediction of 28-day mortality was created and validated. The diagnostic accuracy of this new score was higher than the diagnostic accuracy of MELD, MELD-Na, ALBI, and NLR score.