Original Articles
 
Comparison of Acute on Chronic Liver Failure Prognostic Scores and Utility of qSOFA in Predicting Short Term Mortality
 
Gowtham Kumar N, Athish Shetty, Balaji Musunuri, Shiran Shetty, C Ganesh Pai, Ganesh Bhat
Department of Gastroenterology, Kasturba Medical College, Manipal Academy of Higher Education, Manipal, India.


Corresponding Author:
Dr Ganesh Bhat
Email: drbhatgs@rediffmail.com

Abstract

Background: Acute on chronic liver failure (ACLF) is a condition precipitated by acute insult in a background of cirrhosis, which has a distinct clinical prognosis. qSOFA (Sequential Organ Failure Assessment), being simple bedside scoring system has been used in sepsis patients to predict the mortality. This study aimed at comparing qSOFA with existing ACLF prognostic scores in predicting the short term mortality.
Methods: In this prospective study, patients admitted to a tertiary care hospital with diagnosis of ACLF according to APASL definition were enrolled. We compared qSOFA scoring system with other scoring system : Chronic Liver Failure Consortium Organ Failure (CLIF-C-OF), Asia Pacific Association for the Study of the Liver-ACLF Research Consortium (AARC), Child-Turcotte-Pugh score ( CTP) and Model for End stage Liver Disease-sodium (MELD-Na) in predicting short term (28 and 90day) mortality. Prognostic score performance was evaluated using receiver operating characteristic (ROC) curves.
Results: Total of 112 (99 male-88.3%) patients with mean age of 46.3years were included in the study. On D1, AARC had the highest sensitivity predicting 90day mortality (0.785) compared to CLIF-C-OF (0.749), MELD-Na (0.776), CTP (0.689) and qSOFA (0.549). Among D7 scores, MELD-Na score had the highest sensitivity (0.841), compared to CLIF-C-OF (0.835), AARC (0.834), CTP (0.808) and qSOFA (0.784).
Conclusion: MELD-Na is better predictor of 90 day mortality among available scores.qSOFAis not better than existing ACLF scores in predicting mortality. However,qSOFA can be used to identify high risk ACLF patients at triage to decide on ICU management. Further study is needed to assess usefulness of qSOFA in ACLF patients with systemic inflammatory response syndrome (SIRS/sepsis).

Introduction

Acute on chronic liver failure (ACLF) is a syndrome characterized by acute decompensation of chronic liver disease with organ failure, which carries high short term mortality. The term ACLF was first coined in 1995 to include acute insult in patients with underlying chronic liver disease.1 ACLF is development of acute liver injury, clinically presenting as jaundice and coagulopathy, complicated within four weeks by ascites and or encephalopathy in a patient with previously diagnosed or undiagnosed chronic liver disease.2,3 Early recognition of this entity is important because development of liver failure and organ dysfunction is faster, and there is a fair chance of recovery of liver function with appropriate intervention. 
Patients with ACLF have high short term mortality, and the risk is directly proportional to the number of organ failures.3 Early assessment with appropriate prognostic score will help in deciding appropriate treatment protocols, including the need for ICU care and liver transplantation.  Currently, prognostic scores like Chronic liver failure consortium sequential organ failure assessment (CLIF-C-OF), Asia Pacific Association for the study of the Liver-ACLF Research Consortium (AARC), Model for End-stage Liver Disease-sodium (MELD-Na) and Child-Turcotte-Pugh score (CTP) are being used to prognosticate these patients with variable success.4,5,6,7  ACLF specific prognostic scores like CLIF-C-OF and AARC have been shown to predict outcome more accurately compared to CTP and MELD-Na.4,5 However, highly sensitive and specific tests to accurately prognosticate these patients are still lacking. Also these prognostic scores require different tests and are difficult to use outside the ICU. Due to a delay in getting requisite data, there may be a delay in the identification of patients requiring ICU care.
A new scoring system called quick sequential organ failure assessment (qSOFA) based on three bedside assessment (each allocated one point): respiratory rate (RR= 22 breath per minute), altered sensorium (Glasgow Coma Scale, GCS<15), and systolic blood pressure (SBP<100mmHg) has been evaluated in sepsis patients.9 A qSOFA score = 2 increases all-cause short term mortality in patients with sepsis.10 In view of the complexity of ACLF prognostic scoring system and qSOFA being a simple bedside score, we performed a prospective study comparing the qSOFA score with other ACLF specific scoring system in predicting the short term (90 days) mortality.

Materials and Methods

Study Design and Settings

In this prospective cohort study, patients admitted to Kasturba Hospital Manipal, a tertiary care hospital in India, who qualified for the diagnosis of ACLF according to the APASL definition were enrolled into the study. The study protocol was reviewed and approved by the hospital ethical committee.

Patients and Data Collection

Inclusion criteria:

Patients admitted to the hospital above 18 years of age and fulfilling the APASL definition of ACLF {presentation with jaundice (serum bilirubin>5mg/dl) and coagulopathy (INR>1.5) with development of ascites and/ or encephalopathy within 4 weeks of onset} were included in the study. 

Exclusion criteria:

Following patients with ACLF diagnosis were excluded - patients diagnosed with other organ failures unrelated to the present ailment which can influence the outcomes of this study, patients diagnosed with HIV/Malignancy and patients who had prior h/o liver decompensation.

Prognostic scores:

Prognostic scores were calculated at D1 and D7 of hospitalization. The CTP score (range, 5-15) was calculated by hepatic encephalopathy, ascites, serum bilirubin, albumin and INR. MELD score (range, 6-40) was calculated with formula: 9.57 x log (creatinine mg/dl) + 3.78 x log (bilirubin mg/dl) + 11.2 x log (INR) + 6.43. MELD-Na was calculated by the formula MELD-Na- [0.025x MELD x (140-Na) +140. CLIF-C-OF score was calculated by using the organ failure calculator provided by the Ef-CLIF foundation on their official website. AARC score was calculated by using the calculator provided by the APASL ACLF Research consortium, and qSOFA was calculated by the presence of respiratory rate (RR= 22 breath per minute), altered sensorium (Glasgow Coma Scale, GCS<15), and systolic blood pressure (SBP<100mmHg).
After obtaining written consent, data were collected prospectively on patient’s demographics, clinical features, laboratory parameters, disease severity, aetiology of the underlying chronic liver disease and the acute insult, and presence of multi-organ dysfunction.  A detailed history about alcohol intake, other drugs, including complementary and alternative medications usage, development of abdominal distension, fever, altered sensorium, upper gastrointestinal bleed and risk factors for viral hepatitis were recorded.  Physical examination findings, including vitals, were recorded in a standard format. Hepatic encephalopathy was graded according to the West Haven system from grade 0 to 4.
Blood samples were collected for all required investigations to calculate the scores, and additional biochemical investigations were done according to the patient’s condition. Q-SOFA, CTP score, MELD- Na, CLIF-C-OF and AARC scores were calculated on Day 1 and Day 7 of admission. The ascitic fluid analysis was done to calculate the serum ascites albumin gradient (SAAG) and to rule out spontaneous bacterial peritonitis (SBP). Upper G.I. (gastrointestinal) endoscopy was done for all patients, both for bleeders and non-bleeders, for assessment of portal hypertension where ever it was feasible. Blood, urine, and ascitic fluid cultures were sent for all patients. Anti HAV IgM, HBsAg, Anti HCV, Anti HEV IgM were sent. In HBsAg positive patient’s samples were sent for HBV DNA levels, HBeAg, Anti HBc IgM and total. In patients positive for HCV antibody, samples were sent for HCV RNA levels and HCV genotyping.  All patients underwent USG abdomen and portal venous doppler study. The following details were recorded- liver span, surface nodularity of the liver, size of the spleen, size of the portal and splenic veins, the presence of portosystemic collaterals, hepatic veins, inferior vena cava (IVC) and presence of ascites. Spontaneous bacterial peritonitis (SBP) was defined by> 250 neutrophils per mm3 or positive ascitic fluid cultures.
All the patients were followed up to 90 days for the primary outcome, i.e., mortality and the time of death from the diagnosis were recorded.

Statistical Analysis

All the data collected were entered on an excel sheet. Mean and median were calculated for appropriate variables. All the variables between survivors and non-survivors were compared. All variables, including age, acute kidney injury (AKI), hepatic encephalopathy, oesophagal varices, rapidity of progression of the disease and other laboratory parameters were analysed with univariate logistic regression and which were significant by univariate analysis were again compared by multivariate logistic regression analysis both for 28 days and 90 days mortality with both Day 1 scores and Day 7 scores.
The ability to discriminate between survivors and non-survivors using the different scoring systems was assessed by the receiver operating characteristic (ROC) curve. Sensitivity, specificity and cut off values for all the scores were obtained, and the population at risk of death was assessed by Kaplan–Meir curve, which was then compared by the log-rank test. The positive predictive value (PPV), negative predictive value (NPV), PLR (positive likelihood ratio), NLR (negative likelihood ratio) were calculated for all the scores for predicting mortality. A p-value of less than 0.05 was taken as significant. Statistical analysis was done by SPSS 16 software.

Results

A total of 112 patients who met the inclusion criteria were included in the study. Out of which male patients were 99 (88.3 %), and female patients were 13 (11.7 %) with a mean age of 46.3 years.  All the patients were divided into two groups based on the primary outcome, i.e., mortality and various baseline parameters and scores between these two groups were compared (Table 1).





Out of 112 patients, 40 patients died over a 28 days time period (35.7%), and 46 patients died over a 90 days time period (41.1 %). The mean age of the patients among the survivors was 47.30 ± 10.89 years and among the non-survivors was 45.00 ± 11.11 years. Among the non-survivors (46), 42 were males, and 4 were females. 
Out of all patients, 81 (72.3 %) were newly diagnosed cases of CLD and 31 (27.67 %) were already diagnosed with CLD without any prior history of decompensation. The mean duration of CLD among these patients was 19.25 months. The mean duration from the onset of jaundice/coagulopathy to the development of ascites and or encephalopathy was 11.4 days. Among all the patients, 17 (15.2%) had mild ascites, 62 (55.4%) had moderate ascites, and 33 (29.5%) had severe ascites. Most of the patients had grade-II encephalopathy (56.3%).  Sixty-nine patients (61.6%) had small varices, and 43 patients (38.4%) had large varices. However, only twenty-two patients (19.6%) had upper GI bleed requiring intervention. 
Alcohol was the common acute insult for liver failure (68 patients, 60.7%), followed by infections (28 patients, 25%), drug induced liver injury (DILI) (18 patients, 16.1%), HEV (7 patients, 6.3%), HBV (6 patients, 5.4%), HAV and Wilsons disease (1patient, 0.9% each), and no cause identified (7 patients, 6.3%).  Among the etiology of underlying chronic liver disease, alcohol was the commonest cause (89 patients, 79.5%), followed by NAFLD (7 patients, 6.3%), HBV (6 patients, 5.4%), AIH (5 patients, 4.5%), cryptogenic (4 patients, 3.6%), HCV (3 patients, 2.7%) and Wilsons (1 patient, 0.9%).
In 28 patients, infection was the cause of acute insult, while 36 patients had signs of sepsis during their illness. Among them, SBP was the commonest infection, followed by respiratory infections, lower limb cellulitis and urinary tract infection. E.coli was the commonest organism grown among culture-positive patients (9 patients, 21%).
All the baseline characteristics among both the groups were compared to predict mortality using univariate regression analysis. All the factors like an acute insult, aetiology of the underlying chronic liver disease, age, gender, duration of CLD, presence of GI bleed, AKI, encephalopathy, qSOFA, CLIF-C-OF, CTP, MELD-NA, AARC, the time interval between jaundice and ascites, infections, Hb, TLC, Neutrophil count, platelets, INR, total bilirubin, albumin, urea, creatinine and sodium were compared. Among these, variables like GI bleed, large esophageal varices, AKI, TLC, neutrophils, total bilirubin, urea, creatinine, sodium, lactate and all prognostic scores were found to predict mortality on univariate logistic regression with a p-value of <0.05.
All these factors were again compared using multivariate logistic regression analysis. Only total bilirubin and serum lactate were significant with a p-value of <0.05. 

Prognostic Accuracy of Scoring Systems

Day 1 scores were compared in all patients. Receiver Operator Characteristic (ROC) curves were calculated and found that the AARC score has the highest sensitivity in predicting 90-day mortality with area under the curve (AUC) of 0.785 (95% CI-0.698-0.872) followed by MELD-Na (0.776, 95% CI-0.686-0.865), CLIF-C-OF (0.749, 95% CI-0.659-0.839), CTP (0.698, 95% CI-0.598-0.799),and q SOFA (0.549, 95% CI-0.438-0.659) (Table 2 and Figure 1).







On comparison of Day 7 scores for predicting mortality, MELD-Na score had the highest sensitivity to predict the 90-day mortality with AUC of 0.841 with CI of 0.764-0.917 followed by CLIF-C-OF (0.835, CI-0.757-0.914), AARC (0.834, CI-0.750-0.918), CTP (0.808, CI-0.722-0.894), qSOFA (0.784, CI-0.69-0.878) which is in contrast with results of day 1 scores (Table 3 and Figure 2).







The mortality increased with an increase in the grade of ACLF according to AARC score or CLIF-C-OF score. The same was found on the Kaplan Meir survival analysis of all the five available scores, which showed high short term mortality among the patients of higher grades of ACLF or higher MELD-Na scores or severe disease as assessed by qSOFA or CTP score, which was tested with Mantel-cox Log-rank test with a significant p-value of <0.05 (Figure 3). Among all the scores used to predict 90-day mortality, although Day-1 scores of AARC, CLIF-C-OF, MELD-Na had similar sensitivity,  Day-7 and overall sensitivity was highest for MELD-Na followed by AARC and CLIF-C-OF. We found that q SOFA can also predict mortality but with a low sensitivity when compared to all other scores with good specificity.

Discussion

In our study, alcohol was the commonest cause of acute insult and aetiology of the underlying chronic liver disease with 90 days mortality of around 41%, which is in tune with recent data from India and the western world.11,12 As the majority of these patients had recent alcohol abuse, the mainstay of treatment remains nutritional support, de-addiction and organ support as definitive treatment options like liver transplantation are not feasible. Early recognition of patients requiring ICU care will help in early initiation of organ support, which may help in the final outcome.  
In our study, 32% of patients had infections, E.coli being the most common organism cultured. Infection is one of the major cause of mortality in ACLF patients. Early detection of sepsis before the development of organ failure (window period) and appropriate therapeutic intervention may improve the outcome.13,2 Recently, qSOFA score has been used in sepsis patients to predict the mortality and need for ICU care. Being a simple bedside scoring system, it has shown good sensitivity and specificity in predicting mortality. Early application of qSOFA at admission to predict sepsis and systemic inflammatory response syndrome (SIRS) and to decide appropriate treatment, including ICU care, may alter the outcome of these patients.14
In our study, qSOFA showed low sensitivity (31%) with high specificity in predicting mortality. Among the patients, if we apply these scores to segregate them into low and high-risk groups, qSOFA identified only 35 patients (31%) as high risk in prognosticating when compared to CLIF-C-OF- 47(42%), AARC-74(66%), CTP-(93%), MELD-Na -100(89%) in increasing order as higher grades of disease. This explains the high sensitivity of MELD, CTP and AARC scores in predicting mortality according to the AUROC curves. Even in Kaplan-Meir survival analysis, according to q SOFA, the median survival in the low-risk group was 24.7 days, and in a high-risk group was 13.9 days falling within 95 % CI.
The other highlight of our study is MELD-Na score did well in predicting 90-day mortality when compared to the AARC and CLIF-C- OF score, except for day one scores where the AARC score did well over MELD-Na. The increase in MELD-Na score (D1- D7) overall is the best predictor of 90-day mortality among all the scores included in our study. These findings were comparable to other studies done to compare the prognostic scores.  In a meta-analysis by Zheng YX et al., which included 26 studies, they found that the MELD-Na score had the most considerable AUROC value, which is comparable to CLIF-C-OF in estimating three-month mortality.15 In another study by Perdigoto DN et al., comparing CTP, MELD-Na and CLIF-C-OF score concluded that MELD-Na is better than the other scores for predicting 90 days mortality.16 However, in a study by Choudhury A et al., comparing the AARC with CLIF-C-OF models it was found that the AARC score is more reliable and outperformed the CLIF-C-OF score.5 In contrast studies by Leao GS et al. and Dhiman et al. concluded that CLIF-C-OF better in predicting short term mortality than other scoring systems.17,18 So universally acceptable single prognostic scoring system is yet to established.
The advantage of the q SOFA score is that it has high specificity, positive predictive value (PPV) and accuracy in predicting mortality among ACLF patients. This difference is because it requires at least two extrahepatic organ failures to keep them in the high-risk category, which have a proven negative impact among these patients. It is also easy to use and can be applied bedside with the availability of basic laboratory parameters. The other conventional scores are more elaborate and include more organ failures along with multiple laboratory parameters, which are more complex and difficult to calculate, which might delay the initiation of treatment. However, AARC score and CLIF-SOFA has been shown to be better scores in terms of sensitivity to predict mortality as they include more organ failures, including liver and coagulation, which are more important in a given patient to determine mortality as the primary insult is in the liver.
The limitation of the q SOFA score is the absence of liver parameters in deciding the prognosis of a patient as the signs of SIRS can be present even in the absence of liver failure. Hence it should be used with caution and may be useful in a known case of cirrhosis who presents with features of SIRS or among patients who are suspected of having liver failure with SIRS. This is very important as identification of patients with a high risk of mortality at the time of admission will not only help in deciding ICU management but also in deciding about referral for early liver transplantation. So this cannot be used as a stand-alone single prognostic score among these patients. It can be used as an initial screening tool, and the patients who are at low risk with this score should be subjected to conventional scores like AARC or CLIF C OF to further prognosticate and plan appropriate treatment.
Moreover, the importance of dynamic assessment of patients with the available scores has been shown in our study, which clearly showed that all scores predicted mortality better when measured on day 7 when compared to at the admission.2,19 We also found that serum bilirubin and lactate, along with the scores measured, can predict the 90-day mortality among ACLF patients. Even though liver transplantation is the only definitive treatment option for critically ill patients, early recognition of high-risk patient and initiation of appropriate treatment may also change the outcome, including the need for liver transplantation.  
Our study is a prospective study with a good number of patients enrolled in a single-center with data up to 90 days follow up. This is one of the first study of its kind including q SOFA and checking its validity among ACLF patients and its comparison with other scores. Our study also has some limitations. As this is a single-center study, these results have to be reproduced in other multi-centric cohort trials before drawing conclusions.  Also we did not analyze the utility of qSOFA in ACLF patients with sepsis/SIRS. 
In conclusion, the present study demonstrates that ACLF carries high three-month mortality, and alcohol is the most common acute and chronic aetiology.  We found that q SOFA can also predict mortality but with a low sensitivity when compared to all other scores with good specificity, PPV and accuracy. The qSOFA can be used triage to identify the ACLF patients at a very high risk of mortality to decide on ICU management, followed by application of other conventional scores at D1 and D7 to predict the mortality. Further studiesare needed to find out the utility of qSOFA in ACLF patients with SIRS/sepsis. 

References
  1. Ohnishi H, Sugihara J, Moriwaki H, Muto Y. Acute-on-chronic liver failure. RyoibetsuShokogunShirizu 1995; (7): 217-219.
  2. Sarin SK, Choudhury A, Sharma MK, Maiwall R, Mahtab MA et al. Acute-on-chronic liver failure: consensus recommendations of the Asian Pacific Association for the study of the Liver (APASL): an update. Hepatol Int 2019; 13: 353-390.
  3. Kumar R, Mehta G, Jalan R. Acute- on- Chronic liver failure. Clin Med (Lond) 2020; 20(5): 501-504.
  4. Jalan R, Saliba F, Pavesi M, Amoros A, Moreau R, et al. Development and validation of a prognostic score to predict mortality in patients with acute-on-chronic liver failure. J Hepatol. 2014; 61: 1038-1047.
  5. Choudhury A, Jindal A, Maiwall R, Sharma MK, Sharma BC, et al. Liver failure determines the outcome in patients of acute-on-chronic liver failure (ACLF): comparison of APASL ACLF research consortium (AARC) ane CLIF-SOFA models. Hepatol Int. 2017; 11(5): 461-471.
  6. Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, et al. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology. 2000; 31(4): 864-871.
  7. Leise MD, Kim WR, Kremers WK, Larson JJ, Benson JT, et al. A revised model for end-stage liver disease optimizes prediction of mortality among patients awaiting liver transplantation. Gastroenterology 2011; 140: 1952-1960.
  8. Child CG, Turcotte JG. Surgery and portal hypertension. In: The Liver and Portal Hypertennsion, Child III CG (Ed), Saunders, Philidephia 1964, 50-64.
  9. Seyemour CW, Liu VX, Iwashyna TJ, Brunkhorst FM, Rea TD, et al. Assessment of Clinical Criteria for Sepsis: For the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis 3). JAMA 2016; 315: 762-774.
  10. Raith EP, Udy AA, Bailey M, McGloughlin S, Maclsaac C, et al. Prognostic Accuracy of the SOFA score, SIRS Criteria and qSOFA Score for In-Hospital Mortality Among Adults With Suspected Infection Admitted to the Intensive Care Unit. JAMA 2017; 317: 290-300.
  11. Shalimar, Saraswat V, Singh SP, Duseja A, Shukla A, et al. Acute-on-chronic liver failure in India: The Indian National Association for Study of the Liver consortium experience. J Gastroenterol Hepatol 2016; 31(10): 1742-1749.
  12. Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, et al. Acute-on-Chronic Liver Failure Is a Distinct Syndrome That Develops in Patients With Acute Decompensation of Cirrhosis. Gastroenterology. 2013; 144: 1426-1437.
  13. Katoonizadeh A, Laleman W, Verslype C, Wilmer A, Maleux G, et al. Early features of acute-on-chronic alcoholic liver failure: a prospective cohort study. Gut 2010; 59(11): 1561-1569.
  14. Rodriguez RM, Greenwood JC, Nuckton TJ, Darger B, Shofer FS, et al. Comparison of qSOFA with current emergency department tools for screening of patients with sepsis for critical illness. Emeg Med j. 2018; 35(6): 350-356.
  15. Zheng YX, Zhong X, Li YJ, Fan XG. Performance of scoring systems to predict mortality of patients with acute-on-chronic liver failure: A systematic review and meta-analysis. J Gastroenterol Hepatol. 2017; 32(10): 1688-1678.
  16. Perdigoto DN, Figueiredo P, Tome L. The Role of the CLIF-C OF and the 2016 MELD in Prognosis of Cirrhosis with and without Acute-on-Chronic Liver Failure. Ann Hepatol. 2019; 18(1): 48-57.
  17. Leao GS, Lunardi FL, Picon RV, Tovo CV, de Matto AA, et al. Acute-on-chronic liver failure: A comparison of three different diagnostic criteria. Ann Hepatol. 2019; 18(2): 373-378.
  18. Dhiman RK, Agrawal S, Gupta T, Duseja A, Chawla Y. Chronic Liver Failure-Sequential Organ Failure Assessment is better than the Asia-Pacific Association for the Study of Liver criteria for defining acute-on-chronic liver failure and predicting outcome. World J Gastroenterol. 2014; 28(40): 14934-14941.
  19. Chan AC, Fan ST, Lo CM, Liu CL, Chan SC, et al. Liver transplantation for acute-on-chronic liver failure. Hepatol Int 2009; 3: 571-581.