key: cord-0841898-exp1fnzv
authors: Kaur, Savneet; Hussain, Sadam; Kolhe, Kailash; Kumar, Guresh; Tripathi, Dinesh M.; Tomar, Arvind; Kale, Pratibha; Narayanan, Ashad; Bihari, Chaggan; Bajpai, Meenu; Maiwall, Rakhi; Gupta, Ekta; Sarin, Shiv K.
title: Elevated plasma ICAM1 levels predict 28-day mortality in cirrhotic patients with COVID-19 or bacterial sepsis
date: 2021-05-08
journal: JHEP Rep
DOI: 10.1016/j.jhepr.2021.100303
sha: 17d97569ea2d93939d40db773816a887f5b56868
doc_id: 841898
cord_uid: exp1fnzv

BACKGROUND AND AIMS: Endothelial injury and dysfunction play a detrimental role in the pathogenesis of infections. Endothelium-related molecules have been reported as potential biomarkers for early diagnosis and/or prognosis of infections. The prognostic value of these biomarkers in patients with liver cirrhosis and infections however remain elusive. METHODS: In this study, we investigated the performance of key soluble endothelial injury biomarkers, including intercellular adhesion molecule 1 (ICAM1), von Willebrand Factor (vWF) antigen, vascular endothelial growth factor receptor 1 (vegfr1), angiopoietin 1 and 2 (Ang1, 2) as mortality predictors in liver cirrhosis patients with severe covid-19 or bacterial sepsis. RESULTS: A total of 66 hospitalized patients [admitted in covid-19 ward or liver intensive care unit (ICU)] were included. 22 patients had covid-19 alone while 20 patients had cirrhosis plus covid-19. 24 patients had liver cirrhosis plus bacterial sepsis. Among cirrhosis patients, most common aetiology was alcohol. ICAM1 was increased (p=0.003) while vegfr1 (p<0.0001) and Ang1 (p<0.0001) were reduced in covid-19 patients with cirrhosis, as compared to covid-19 patients alone. Endothelial biomarkers between cirrhosis patients with severe covid-19 or with bacterial sepsis in the ICUs did not differ significantly. In these patients, ICAM1 levels significantly and independently predicted mortality (HR:3.24;1.19-8.86) along with MELD, renal and coagulation failures. The area under the curve for ICAM1 was 0.74, MELD was 0.60 and combined ICAM1 and MELD was 0.70. ICAM1 also positively correlated with the composite organ failure scores recorded 3-5 days post ICU admission (CLIF-OF and SOFA) in this subgroup of patients. CONCLUSION: The study indicates that in patients with liver cirrhosis, elevated plasma ICAM1 serve as an independent predictor of severe covid-19 or sepsis associated 28-day mortality. LAY SUMMARY: Bacterial sepsis and covid-19 lead to increased mortality in cirrhosis patients. In this study, we demonstrate that high plasma levels of ICAM1, an endothelial injury biomarker is one of the important factors predicting mortality in critically-ill cirrhotic patients having severe covid-19 or bacterial sepsis.

The study was supported partly by research funds from the Institute of liver and biliary Sciences (ILBS), New Delhi and partly by Department of Science and Technology (DST), Government of India.

Introduction Endothelium is a highly specialized and dynamic organ that regulates several processes in both physiology and pathophysiology. 1 It is the first organ to discern and mount compensatory responses to any injury signals, thus acting as a shield to protect the underlying tissue. An unhealthy endothelium is associated with profound changes in many physiological systems including expression of adhesion molecules, maintenance of adequate blood vessel tone and haemostasis. Endothelial activation/injury is thought to be a key event in the development and pathogenesis of distinct human vascular diseases, including liver cirrhosis, hypertension, diabetes, atherosclerosis, sepsis and now even covid-19. 2 Patients with covid-19 exhibit attributes correlating with endothelial dysfunction involving changes in vascular permeability, inflammation, accumulation and extravasation of leucocytes, activation of procoagulant pathways and disruption of alveolar-capillary barrier. 3, 4 A high mortality rate and poor clinical outcomes due to covid-19 infections in elderly subjects and patients with comorbid conditions including diabetes, obesity and hypertension has also been postulated to be because of an underlying endothelial dysfunction. 5 In patients with cirrhosis, an infection with SARS-CoV-2 leads to a rapid clinical deterioration in otherwise stable patients, leading to increased mortality. 6, 7 Since biomarkers of endothelial activation and dysfunction are associated with poor outcomes in patients with both covid-19 and sepsis, we studied these biomarkers in patients with liver cirrhosis having either covid-19 or bacterial sepsis. Plasma levels of some of the well-known biomarkers of endothelial injury, including intercellular adhesion molecule 1 (ICAM1), von Willebrand Factor (vWF) antigen, vascular endothelial growth factor receptor 1 (vegfr1), angiopoietin 1 and 2 (Ang1, 2) were studied for their association with 28-day mortality in patients with covid-19 with and without liver cirrhosis as well as in those with sepsis plus liver cirrhosis.

Covid-19 patients (≥18 years) were selected based on a confirmed diagnosis on PCR assays (two viral genes, envelope protein or E gene and RNA dependent RNA polymerase or RdRP gene) of nasopharyngeal swab samples. Covid-19 patients were categorised as mild, moderate and severe covid-19 infection. 9 Patients with liver cirrhosis plus bacterial sepsis hospitalized in the intensive care unit (ICU), ILBS were also recruited in the study. Sepsis diagnosis included an increase in the SOFA score of 2 points (baseline SOFA) or more as a result of documented infection. 10 Some samples were also collected from non-hospitalised, mild and asymptomatic covid-19 patients from hospital staff. Patient groups included in the study are given in Supp Figure 1 . Patients post-liver transplant or transplanted during hospital stay and also with any malignancy were excluded.

The ILBS ethics committee approved this study (Ethics protocol no IEC/2020/82/NA01).

Samples from biobank were collected as per the ILBS biobank guidelines, and samples from in-house hospitalized patients and hospital staff were obtained after proper informed consent from the patient or his/her relative following the national bioethical guidelines. Commercial 

Categorical variables were expressed as percentages while continuous variables were expressed either as man (SD) for continuous distribution or as median (minimum-maximum) for variables for skewed distribution. Categorical data were analysed with Chi-square test or

Fisher's exact test wherever appropriate. Continuous variables were compared with independent sample t-test or MannWhitney U test for two groups and by One-way Anova or J o u r n a l P r e -p r o o f Kruskal-Wallis test for more than two groups depending on the distribution. Differences between endothelial cell markers among various groups and also between survivors and nonsurvivors was calculated by using the unpaired two-tailed MannWhitney U test. Survivors were those alive or discharged within 28 days. Clinical predictors were entered to the Cox hazard regression model to assess the effects of factors on 28-day in-hospital mortality. We used unadjusted linear regression models and multivariate models adjusted for clinical covariates. The receiver operating characteristic (ROC) curves were plotted to explore the area under the curve (AUC). Cut-off points to discriminate the survivors from the nonsurvivors were calculated by obtaining the best Youden index (sensitivity% + specificity% − 100). The cut-off values were taken where sensitivity and specificity were optimal. AUCs were expressed with their 95% confidence interval (CI). The study participants were divided into 2 groups according to the optimal cut-off value of the probability of 28-day mortality, and then the 28-day mortality rates of the groups were compared using Kaplan-Meier curve analysis. A log-rank test was conducted to compare the survival curves of the groups. 

A total of 42 hospitalized covid-19 patients (admitted in dedicated covid-19 wards or ICUs) were included in the first analysis ( Table 1) . Out of these, 22 were in covid-19 category and 20 were in the covid-19 plus liver cirrhosis category. We also included asymptomatic nonhospitalized patients (aged 34-61y) with mild covid-19 as controls (n=6) as a comparator group for endothelial biomarker levels in covid-19 patients without cirrhosis. In the hospitalized patients, there was no significant difference among the groups with respect to age and sex. In patients of liver cirrhosis with covid-19, most common aetiology of the liver disease was alcohol with 50% in this category followed by non-alcoholic steatohepatitis (NASH). Most of the cirrhosis patients (65%) belonged to CTP class C. Ascites presenting as decompensating event was observed in 70% and hepatic encephalopathy was observed in 40% of covid-19 patients with liver cirrhosis. The presence of ARDS was similar in patients of both the groups.

As far as comorbidities were concerned, diabetes was the most common in both covid-19 and covid-19 with liver cirrhosis groups (50% in covid-19 and 45% in covid-19 plus liver cirrhosis groups respectively). The charlson comorbidity index or CCI was significantly different between covid-19 plus liver disease groups [1 (0-5) Table 2 ). Next we studied these biomarkers in patients with liver cirrhosis and covid-19. Patients with covid-19 plus liver cirrhosis had significantly higher levels of ICAM1 (p=0.003) while those of vegfr1(p<0.0001) and Ang1 (p<0.0001) were substantially reduced in comparison to that observed in patients with covid-19 alone ( Table 2 ). The levels of vWF and Ang2 were not significantly different between patients with covid-19 alone and those with covid-19 and liver cirrhosis ( Table 2) .

Endothelium being a key component involved the development of macro-and microcirculatory disturbances and organ dysfunction in infections in severely-ill ICU patients, we next studied these biomarkers in a subgroup of ICU patients with severe covid-19 and those with severe covid-19 plus cirrhosis. We also included ICU admitted liver cirrhosis patients with sepsis as one of the groups here (Table 3 ). In addition to usual clinical parameters, factors including Ferritin, LDH, PCT, D-dimers etc that are known to be associated with clinical outcomes in severe covid-19 and bacterial sepsis in ICUs were recorded for these patients (Table 3) . Among these three categories, platelet counts, bilirubin levels, albumin, INR, creatinine, ferritin levels and CCI score were significantly different.

Alcohol was the major aetiology and diabetes as major comorbidity in sepsis patients with liver cirrhosis similar to covid-19 with cirrhosis patients. In sepsis plus cirrhosis group, J o u r n a l P r e -p r o o f ascites with present in 95% and hepatic encephalopathy was present in 66.6% patients, which was not different in comparison to that seen in cirrhosis plus covid-19 patients. Similar to cirrhosis plus covid-19 group, most of the sepsis patients were also in CTP class C category but their MELD scores were higher than covid-19 plus cirrhosis groups (p=0.01). Among the three categories, the number of patients with mechanical ventilation was lowest in covid-19 alone category while comparable in liver cirrhosis plus or covid-19 groups (Table 3) . With respect to SOFA, CLIF-OF scores and individual organ failures in cirrhotic patients at day 3-5 in ICU, the scores were comparable between patients with covid-19 infection and those with sepsis. In terms of biomarker levels, cirrhosis patients with covid-19 or bacterial sepsis respectively showed significantly higher levels of ICAM1 (p=0.04, 0.001) and reduced levels of vegfr1 (p=0.003, 0.007) and Ang1 (p=0.0009, 0.001) as compared to patients with severe covid-19 alone. Intriguingly, we did not observe any significant differences in the levels of endothelial biomarkers, ICAM1, vWF, vegfr1, Ang1 and Ang2 between cirrhosis patients with sepsis and those with covid-19 ( Figure 1 ).

We estimated the 28-day survival in the three different categories of severely-ill ICU patients. Severe covid-19 patients showed a survival of 72.7% while severe covid-19 with liver cirrhosis and those with sepsis and liver disease had a survival of 26.7% and 19.8% respectively at 28 days ( Figure 2 ). Log-rank analysis revealed a significant difference between the survival of patients with sepsis and liver cirrhosis in comparison to patients with severe covid-19 alone (chi square=4.03, df=1, p=0.04). Median time of survival in severe covid-19 was 16 days (95% CI 10.33-21.66), in severe covid-19 with liver cirrhosis was also 16 days (95% CI 7.51-24.48) and in sepsis plus liver cirrhosis , it was 9 days (95% CI 4.70-13.2). In patients of liver cirrhosis with sepsis, mortality was due to severe sepsis or septic J o u r n a l P r e -p r o o f shock with progressive liver failure or multi-organ failure. We studied if the endothelial biomarkers predicted survival in these severe ICU patients'. In the comparison between nonsurvivors and survivors in this group, the levels of ICAM1 (p<0.0001) were significantly elevated while that of vegfr1 (p=0.04) and Ang1 (p=0.02) were decreased in the nonsurvivors as compared to the survivors (Supp Table 3 ).

We next studied these biomarkers as mortality predictors along with other clinical biomarkers in the cohort of ICU patients using logistic regression. The association between logtransformed biomarkers and death was analysed. Unadjusted univariate and adjusted hazard ratios of all the clinical variables derived from the Cox hazards model in all ICU patients are given in Supp Table 4 . In the univariate analysis, D-dimers, AST, Creatinine and ICAM1 were significant predictors of mortality in the full cohort of ICU patients. AST (HR=1.87 95% CI 1.08-3.31, p=0.03) and ICAM1 (HR=2.94 95% CI 1.3-6.65, p=0.01) emerged as significant and independent predictors of mortality after multivariate analysis (Supp Table 4 (Table 4 ). When CLIF-OF and SOFA composite scores were included in the multivariate analysis along with other significant predictors, they only appeared as independent mortality predictors without ICAM1 (Table 4) .

The comparison between discriminatory values of ICAM1 and various scoring systems was performed in liver Cirrhosis Patients with severe Covid-19 or Sepsis (Supp Table 5 ). An AUC of 0.83 (95% CI 0.67-0.97) was obtained for CLIF-SOFA and for SOFA score, it was 0.82 (95% CI 0.66-0.98). In case of ICAM1, the AUC was 0.75 (95% CI 0.56-0.93) while that for MELD, it was 0.60 (95% CI 0.38-0.83, Figure 3A ).We then evaluated the sensitivity and specificity of prediction at cut-off points that provided the best Youden index. The sensitivity and specificity of ICAM1 was comparable to the organ failure scores at the respective cut-off values but higher than the MELD score. The best Youden indices were obtained for the organ failure scores and ICAM1 as compared to MELD score (Supp Table   5 ). Log ICAM1 levels above 12.3 was the cut-off value with sensitivity of 72%, specificity of 72.7% and Youden index of 44.7% (AUC 0.74, 95% CI 0.56 -0.93, Figure 3A ). With the obtained cut-off values of ICAM1, we next constructed survival curves in patients with liver cirrhosis using the Cox regression. The log-rank analysis showed that mortality in patients with logICAM1 value less than or equal to 12.3 was significantly lower at 28 days than patients with logICAM1 levels greater than 12.3 (Chi-Square: 5.10, p=0.02). The HR at this cut-off value of ICAM1 was 3.24 with 95% CI 1.43-7.38, p=0.005 ( Figure 3B ). The cut-off value of MELD came out to be 29.5 with sensitivity of 60%, specificity of 63.6% and Youden index of 23.6%. Using this cut-off values of MELD score, we studied survival curves in patients with liver cirrhosis. The analysis showed a significant difference in mortality in J o u r n a l P r e -p r o o f this study cohort using this cut-off of MELD score (Chi-Square: 6.1, p=0.013). The HR at this cut-off value of MELD was 2.91 with 95% CI 1.35-6.23, p=0.006 ( Figure 3C ). Overall, ICAM1 showed a net benefit of 14% than MELD alone in discriminating the survivors from non-survivors in this category of patients. Using a combined model of ICAM1 and MELD in liver disease patients with severe covid-19 or sepsis, the AUC value to predict 28-day mortality was 0.70, 95% CI 0.49 -0.91, demonstrating a net benefit of 10% than that observed with MELD alone ( Figure 3A ).

We next analysed the correlation between the plasma ICAM1 levels and some of the known prognostic factors of critical disease including age, CCI score, CLIF-OF and SOFA scores (3- 11 Ang2/Ang1 ratios have also been shown to be a relevant predictive factor for ICU admission in covid-19 patients. 12 In our study, we observed significantly enhanced levels of in ICAM1 while decreased levels of vegfr1 and Ang1 levels among the survivors and non-survivors in the ICU patients. Our findings corroborate previous findings that increased expression of endothelial cell adhesion molecules is related to covid-19 disease severity. 13, 14 Intriguingly, we found that the soluble ICAM1 levels were further elevated while vegfr1 and Ang1 were decreased in age and sex matched patients having covid-19 plus liver disease as compared to those with covid-19 alone. Since we have not performed a longitudinal analysis of ICAM1-1 in patients before and after covid-19 diagnosis, we can't say that these levels have increased due to SARS-CoV2 infection. Indeed, ICAM1 and VCAM1 are also highly expressed on inflamed liver sinusoidal endothelial cells and circulating ICAM1 levels and even other biomarkers of endothelial injury such as Ang2 have been reported to be significantly elevated in patients with cirrhosis as compared to normal healthy subject. [15] [16] [17] [18] Thus, patients with liver cirrhosis might present with pre-existing altered levels of some of these biomarkers due to an ongoing endothelial injury and activation, In our cohort, liver disease patients with covid-19 had a higher CCI in comparison to those with covid-19 alone.

The presence of comorbidities has been appreciably correlated with endothelial injury in previous reports. 1 Among all the clinical variables in our study, high AST, creatinine, D-dimers and soluble ICAM1 plasma levels emerged as an important predictors of 28-day mortality in ICU patients with AST and ICAM1 appearing as independent predictors after multivariate analysis.

Elevated D-dimers have been earlier reported to be one of the strong predictors of severe prognosis in covid-19 patients. 19 AST and ICAM1 levels also predicted mortality independently in liver cirrhosis ICU patients with either with covid-19 or sepsis. In terms of ROC, we observed that ICAM1 alone or a combined ICAM1 and MELD model was 10% better predictor of short-term mortality as compared to MELD score alone in liver cirrhosis patients with severe covid-19 or bacterial sepsis underscoring the relevance of endothelial injury as one of the important prognostic factor in mortality-related to bacterial or viral infections in patients with liver cirrhosis. A higher mortality in liver cirrhosis patients infected with covid-19 as compared to patients with cirrhosis alone, in spite of low MELD has been documented. Also, in acute on chronic liver failure (ACLF) patients with sepsis and ≥1 organ failure(s), it has been recently reported that MELD-Na does not correctly capture 90-day mortality risk 20 . Our findings reinforce these findings that prognosis and survival could not be only explained by differences in the severity of liver disease. 6 Although not the sole component of host response to infections, endothelium is a key perpetuator of inflammation and coagulation that leads to microvascular occlusion, hypoxia, and organ dysfunction. ICAM1 expressed on the endothelial cells is involved mainly in leukocyte recruitment to the underlying tissues. Enhanced expression of ICAM1 on activated endothelial cells has been reported to contribute to increased neutrophil adhesion and J o u r n a l P r e -p r o o f transmigration into the tissues in inflammation. 21 The activation of neutrophils has been linked to platelet activation and intravascular coagulation, thus causing impaired organ perfusion driving its failure. High soluble ICAM1 levels have been well correlated with the development of multiorgan dysfunction in patients with severe sepsis. 22 We also observed a significant positive correlation between high ICAM1 levels and organ failure severity (both SOFA score and CLIF-OF score). As compared to other liver disease severity markers such as APACHE II and CTP scores, studies have shown that the organ failure scores perform far better in terms of their ability to predict survival in critically-ill cirrhosis patients in the ICUs. 23 In fact, when we entered the CLIF-OF or SOFA scores in our regression models, ICAM1 did not appear as an independent predictor given its close correlation with organ failures scores. When we used the individual organ failures in our statistical analysis, along with ICAM1, renal and coagulation failures appeared as independent mortality predictors in our study cohort similar to a recently published study. 24 Another interesting observation of our study was that in patients with liver cirrhosis, most of the endothelial dysfunction biomarkers were not different in either sepsis or severe covid-19, implying that both covid-19 infection and bacterial sepsis involve considerably similar mechanisms of vascular injury and organ dysfunction. Loss of endothelial barrier function and increased permeability is already well-established in sepsis. Serum concentrations of several endothelial injury biomarkers correlate with sepsis severity and patient mortality. 25 In fact, liver cirrhosis patients with severe covid-19 or sepsis were also very similar in their baseline clinical profile, suggestive of similar ongoing host response pathophysiology in these infections. Several studies have now emphasized that critically-ill patients with sepsis or severe covid-19 have similar immune profile and more than hyperinflammation, it is the immune suppression that is responsible for mortality in these patients. 26 Severe depletion of J o u r n a l P r e -p r o o f immune effector cells and lymphopenia has been unequivocally reported to be a strong predictor of prognosis in both sepsis and covid-19 patients. 26, 27 Aberrations in the innate and adaptive components of the immune system also characterize patients with decompensated liver cirrhosis, whereby cirrhosis-associated immune dysfunction phenotype switches to the predominantly "immunodeficient" one, leading to sepsis and mortality. 28 Although an increased expression of endothelial ICAM1 is correlated with inflammation and neutrophil transmigration in the tissues, it is known that as inflammatory responses progress, very high levels of soluble isoforms of the adhesion molecules that are shed from the endothelial cells serve regulatory and immunosuppressive roles to dampen inflammation (and specifically to reduce leucocyte-endothelial interactions) and protect the host from excessive collateral damage. 29 An elegant study has depicted that soluble ICAM1 shed into the vascular compartment may lead to a self-limiting mechanism of the inflammatory response and hence immune-suppression while the presence of ICAM1 into the distal airway compartment could propagate the already existing inflammation due to enhanced production of cytokines by lung macrophages. 30 How do soluble levels of ICAM1 cause immune-suppression during severe infections is a crucial way forward for further investigations. One of the important limitations of our study is the small sample size. Also we have not assessed serial levels of these markers particularly in resolved phase of the disease in this study. However, to our knowledge, this is the first study that has demonstrated the role of dysfunctional endothelium in covid-19 or sepsis associated mortality in cirrhotic patients. Since the study has been performed in the Asia Pacific region, where we follow the APASL definition for ACLF 31 , none of the decompensated cirrhotic patients in our cohort could be characterized as ACLF according to this criteria. However given the fact that organ failures are significant predictors of mortality in ACLF patients, it would be worthwhile to study the association of ICAM1 levels and other vascular injury biomarkers with mortality in ACLF patients as specified by the EASL as well as APASL guidelines. 20, 23, 31 J o u r n a l P r e -p r o o f 

Alcohol N (%) 0 (0) 6 (50) 10 (41. 

Physiology of the endothelium

The vascular endothelium and human diseases

The procoagulant pattern of patients with COVID-19 acute respiratory distress syndrome

Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in Covid-19

The enigma of endothelium in COVID-19

High rates of 30-day mortality in patients with cirrhosis and COVID-19

Outcomes following SARS-CoV-2 infection in patients with chronic liver disease: an international registry study

Liver sinusoidal endothelial cells: Physiology and role in liver diseases

Ministry of family health and welfare. Guidelines on clinical management of COVID-19

The SOFA score-development, utility and challenges of accurate assessment in clinical trials

Endotheliopathy in COVID-19-associated coagulopathy: evidence from a singlecentre, cross-sectional study

Angiopoietin-2 as a marker of endothelial activation is a good predictor factor for intensive care unit admission of COVID-19 patients

Elevated Expression of Serum Endothelial Cell Adhesion Molecules in COVID-19 Patients

Endothelial Dysfunction and Thrombosis in Patients With COVID-19-Brief Report

The liver: a model of organ-specific lymphocyte recruitment

Circulating ICAM1 in alcoholic liver cirrhosis

Angiogenesis-related biomarkers in patients with alcoholic liver disease: their association with liver disease complications and outcome

Angiopoietin-2/angiopoietin-1 as non-invasive biomarker of cirrhosis in chronic hepatitis C

Clinical and laboratory features of COVID-19: Predictors of severe prognosis

Model for end-stage liver disease-sodium underestimates 90-day mortality risk in patients with acute-onchronic liver failure

The pulmonary endothelial glycocalyx regulates neutrophil adhesion and lung injury during experimental sepsis

Endothelial adhesion molecules and multiple organ failure in patients with severe sepsis

Acute-on-chronic liver failure: Objective admission and support criteria in the intensive care unit

Renal and brain failure predict mortality of patients with acute-on-chronic liver failure admitted to the intensive care unit

The effects of sepsis on endothelium and clinical implications

Severe immunosuppression and not a cytokine storm characterizes COVID-19 infections

Lymphopenia in critically ill COVID-19 patients: A predictor factor of severity and mortality

Cirrhosis-associated immune dysfunction: distinctive features and clinical relevance

Elevated soluble ICAM1 levels induce immune deficiency and increase adiposity in mice

Soluble ICAM1 activates lung macrophages and enhances lung injury

APASL ACLF Research Consortium (AARC) for APASL ACLF working Party

Acute-on-chronic liver failure: consensus recommendations of the Asian Pacific association for the study of the liver (APASL): an update

SD) or Median (min-max) and 'P' values have been calculated by Student's 't' test or MannWhitney U test. Categorical variables are given as N (%) and 'P' values for categorical variables has been calculated using Fisher exact test. ARDS: Acute Respiratory Distress Syndrome; HBV: Hepatitis B Virus; HCV: HepatitisC Virus; NASH:Non-alcoholic steatohepatitis

Multivariate Analysis (with inclusion of composite scores without individual organ failures) SOFA

Significance was taken as p<0.05. HR: Hazard Ratio; CI: Confidence Interval. CCI: Charlson comorbidity index; LDH: Lactate dehydrogenase; NLR: Neutophil to lymphocyte ratio; PCT: Procalcitonin; MELD: Model for end-stage liver disease

We would like to thank all the supporting hospital staff who have helped us with the collection of blood samples.J o u r n a l P r e -p r o o f