Severity of systemic inflammation is the main predictor of ACLF and bleeding in individuals with acutely decompensated cirrhosis

Abstract

Background & aims: Hypercoagulability and hypofibrinolysis in acutely decompensated cirrhosis (AD) may be implicated in disease progression and haemostatic complications. We conducted a prospective study to: (1) characterise haemostatic alterations in AD; (2) evaluate whether such alterations can predict acute-on-chronic liver failure (ACLF) and bleeding/thrombosis.

Methods: Hospitalised individuals with AD were prospectively recruited and underwent an extensive haemostatic profiling including coagulation factors, thrombomodulin-modified thrombin generation assay with evaluation of endogenous thrombin potential (ETP; marker for plasmatic hypercoagulability), fibrinolytic factors, and plasmin-antiplasmin complex (fibrinolysis activation marker). Inflammation severity was assessed by C-reactive protein (CRP). In part 1 of the study, we compared haemostasis in AD vs. controls (stable decompensated and compensated cirrhosis). In part 2 of the study, we prospectively followed individuals with AD for 1 year and investigated predictors of ACLF and bleeding/thrombosis.

Results: A total of 169 individuals with AD were recruited (median model for end-stage liver disease score 20; CLIF-C AD 54). Compared with controls, AD was associated with more pronounced hypercoagulability (ETP: 871 vs. 750 vs. 605 nmol/L per min; p <0.0001), without differences in fibrinolysis activation. During follow-up, 55 individuals developed ACLF. CLIF-C AD, CRP, and Child-Pugh were independently associated with ACLF. A predictive model combining these variables (Padua model) accurately identified individuals at higher risk of ACLF (AUROC 0.857; 95% CI 0.798-0.915; sensitivity 74.5%, specificity 83.3%). Notably, CRP and progression to ACLF, but not baseline coagulopathy, were associated with bleeding (n = 11); CRP and antifibrinolytic factor PAI-1 >50 ng/ml were associated with thrombosis (n = 14). The prognostic value of the Padua model was validated in an independent, bicentric European cohort (N = 301).

Conclusions: Inflammation severity, and not coagulopathy, is the most important predictor of ACLF and bleeding in AD. The Padua model can be used to identify individuals with AD at risk of ACLF.

Impact and implications: A better understanding of haemostasis in individuals with acutely decompensated cirrhosis may help to identify those at higher risk of progression and complications. In this prospective study, we found no significant association between alterations of haemostasis and cirrhosis progression, indicating that the assessment of haemostatic alterations is not useful to identify those at risk. However, we found that C-reactive protein (a simple blood test that reflects severity of inflammation) and severity of chronic liver disease itself (as assessed by specific scores) were associated with cirrhosis progression and development of bleeding complications. Therefore, we developed a simple predictive model - based on C-reactive protein and liver disease scores - that, if validated by independent studies, could be used in clinical practice to assist physicians in identifying individuals with decompensated cirrhosis at higher risk of disease progression and death (i.e. in whom to consider an expedited evaluation for liver transplantation).

Keywords: Acute-on-chronic liver failure; Bleeding; Coagulation; Fibrinolysis; Thrombosis.