Comparison of liver fibrosis scores for predicting mortality and morbidity in heart failure with preserved ejection fraction

Abstract

Aims: Liver fibrosis scores (LFSs) are non-invasive and effective tools for estimating cardiovascular risks. To better understand the advantages and limitations of currently available LFSs, we determined to compare the predictive values of LFSs in heart failure with preserved ejection fraction (HFpEF) for primary composite outcome, atrial fibrillation (AF), and other clinical outcomes.

Methods and results: This was a secondary analysis of the TOPCAT trial, and 3212 HFpEF patients were enrolled. Five LFSs, namely, non-alcoholic fatty liver disease fibrosis score (NFS), fibrosis-4 score (FIB-4), BARD, aspartate aminotransferase (AST)/alanine aminotransferase (ALT) ratio, and Health Utilities Index (HUI) scores were adopted. Cox proportional hazard model and competing risk regression model were performed to assess the associations between LFSs and outcomes. The discriminatory power of each LFS was evaluated by calculating the area under the curves (AUCs). During a median follow-up of 3.3 years, a 1-point increase in the NFS [hazard ratio (HR) 1.10; 95% confidence interval (CI) 1.04-1.17], BARD (HR 1.19; 95% CI 1.10-1.30), and HUI (HR 1.44; 95% CI 1.09-1.89) scores was associated with an increased risk of primary outcome. Patients with high levels of NFS (HR 1.63; 95% CI 1.26-2.13), BARD (HR 1.64; 95% CI 1.25-2.15), AST/ALT ratio (HR 1.30; 95% CI 1.05-1.60), and HUI (HR 1.25; 95% CI 1.02-1.53) were at an increased risk of primary outcome. Subjects who developed AF were more likely to have high NFS (HR 2.21; 95% CI 1.13-4.32). High levels of NFS and HUI scores were a significant predictor of any hospitalization and hospitalization for heart failure. The AUCs for the NFS in predicting primary outcome (0.672; 95% CI 0.642-0.702) and incident of AF (0.678; 95% CI 0.622-0.734) were higher than other LFSs.

Conclusions: In light of these findings, NFS appears to have superior predictive and prognostic utility compared with AST/ALT ratio, FIB-4, BARD, and HUI scores.

Clinical trial registration: (https://clinicaltrials.gov). Unique identifier: NCT00094302.

Keywords: Adverse outcomes; Heart failure with preserved ejection fraction; Liver fibrosis score.

Figure 1

Kaplan–Meier curves for cumulative events of primary composite outcome during follow‐up according to the liver fibrosis scores risk strata. (A) non‐alcoholic fatty liver disease fibrosis score; (B) fibrosis‐4 score; (C) diabetes mellitus score: (D) aspartate aminotransferase/alanine aminotransferase ratio; (E) Health Utilities Index.

Figure 2

Receiver operating characteristic (ROC) curves at 5 years for the LFSs in predicting (A) primary composite outcome, (B) HF hospitalization, (C) any hospitalization, (D) all‐cause death, (E) cardiovascular death during follow‐up. AUC, area under the curve; AST/ALT, aspartate aminotransferase/alanine aminotransferase ratio; FIB‐4, fibrosis‐4 score; HUI, Health Utilities Index; NFS, non‐alcoholic fatty liver disease fibrosis score.

Figure 3

Receiver operating characteristic (ROC) curves at 5 years for the LFSs in predicting incident atrial fibrillation during follow‐up. AUC, area under the curve; AST/ALT, aspartate aminotransferase/alanine aminotransferase ratio; FIB‐4, fibrosis‐4 score; HUI, Health Utilities Index; NFS, non‐alcoholic fatty liver disease fibrosis score.