Incidence and Bedside Predictors of the First Episode of Overt Hepatic Encephalopathy in Patients With Cirrhosis

Abstract

Introduction: Hepatic encephalopathy (HE) is associated with marked increases in morbidity and mortality for patients with cirrhosis. We aimed to determine the risk of and predictors for HE in contemporary patients.

Methods: We prospectively enrolled 294 subjects with Child A-B (70% Child A) cirrhosis and portal hypertension without previous HE from July 2016 to August 2018. The primary outcome was the development of overt HE (grade >2). We assessed the predictive power of model for end-stage liver disease-sodium (MELD-Na) score, the Inhibitory Control Test, the Sickness Impact Profile score, and the Bilirubin-Albumin-Beta-Blocker-Statin score. We also derived a novel predictive model incorporating MELD-Na score, impact of cirrhosis on daily activity (Likert 1-9), frailty (chair-stands per 30 seconds), and health-related quality of life (Short-Form 8, 0-100).

Results: The cohort's median age was 60 years, 56% were men, and the median MELD-Na score was 9. During a follow-up of 548 ± 281 days, 62 (21%) had incident overt HE with 1-year probability of 14% ± 2%, 10% ± 2%, and 25% ± 5% for Child A and B. The best model for predicting the risk of overt HE included MELD-Na, Short-Form 8, impact on activity rating, and chair-stands within 30 seconds. This model-MELDNa-Actvity-Chairstands-Quality of Life Hepatic Encephalopathy Score-offered an area under the receiver operating curve (AUROC) for HE development at 12 months of 0.82 compared with 0.55, 0.61, 0.70, and 0.72 for the Inhibitory Control Test, Sickness Impact Profile, Bilirubin-Albumin-Beta-Blocker-Statin, and MELD-Na, respectively. The AUROC for HE-related hospitalization was 0.92.

Discussion: This study provides the incidence of HE in a well-characterized cohort of contemporary patients. Bedside measures such as activity, quality of life, and physical function accurately stratified the patient's risk for overt HE.

Figure 1:. Incidence and Risk of Overt Hepatic Encephalopathy (HE)

A: Cumulative Incidence of Overt HE B. Cumulative Incidence of Hospitalization for Overt HE Cumulative incidence of the risk of overt HE accounting for the competing risk of death or transplantation in patients with Child A vs. Child B cirrhosis.

Figure 1:. Incidence and Risk of Overt Hepatic Encephalopathy (HE)

A: Cumulative Incidence of Overt HE B. Cumulative Incidence of Hospitalization for Overt HE Cumulative incidence of the risk of overt HE accounting for the competing risk of death or transplantation in patients with Child A vs. Child B cirrhosis.

Figure 2:

Classification Tree for the 1-year Probability of Overt Hepatic Encephalopathy (HE) The important branch-points for each of the model components are illustrated with their corresponding 1-year cumulative risk of Hepatic Encephalopathy (HE) at the bottom. MELD-Na (Model for Endstage Liver Disease-Sodium), SF-8 (Short-form 8, scored 0–100), WPAI (Work-Productivity-Activity Index, scored 0–10). Two examples are provided using boldin to highlight how the lowest and highest risk patients can be classified on the basis of the model components.

Figure 3:. Reclassification of Overt Hepatic Encephalopathy (HE) Risk Relative to MELD-Na

We used two methods to determine the degree to which each model reclassifies the risk of overt HE relative to MELD-Na. These data capture the proportion of patients who were reclassified for their risk of developing or not developing HE using BABS score (bilirubin-albumin-beta-blocker-statin), SIP (Sickness Impact Profile) score, ICT (inhibitory control test) lures, and MASQ-HE score, relative to MELD-Na. A: Integrated discrimination improvement (IDI). Values >0 indicates improved discrimination. Only MASQ-HE improves risk discrimination B: The IDI is based on the relative improvement in discrimination of both developing and not developing the outcome. This panel demonstrates that MASQ-HE performs well by identifying persons at-risk for HE that would not otherwise be identified by MELD-Na. The other models do not improve and may actually worsen risk discrimination C: Category-free net reclassification index (NRI). Values >0 indicates improved discrimination. Only MASQ-HE improves risk discrimination. D: NRI is based on the proportion of patients who are reclassified by each model for their risk of developing or not developing HE. The result is an absolute proportion. Notably, all models change the classification of risks relative to MELD-Na. The NRI does not speak to the correctness of this classification and therefore must be taken into context with other measures such as the c-statistic.