Using liver stiffness to predict and monitor the risk of decompensation and mortality in patients with alcohol-related liver disease

Background & Aims Liver stiffness measurements (LSM) are recommended for disease prognostication and monitoring. We evaluated if LSM, using transient elastography, and LSM changes predict decompensation and mortality in patients with alcohol-related liver disease (ALD). Methods Observational cohort study of compensated patients at risk of ALD from Denmark and Austria. We evaluated the risk of decompensation and all-cause mortality, stratified for compensated advanced chronic liver disease (cACLD: Baseline LSM ≥10 kPa) and LSM changes after a median of 2 years. In patients with cACLD, we defined LSM changes as (A) LSM increase ≥20% ("cACLD increasers") and (B) follow-up LSM <10 kPa or <20 kPa with LSM decrease ≥20% ("cACLD decreasers"). In patients without cACLD, we defined follow-up LSM ≥10 kPa as a LSM increase ("No cACLD increasers"). Remaining patients were considered LSM stable. Results We followed 536 patients for 3,008 patient-years, median age 57 years (IQR 49–63), baseline LSM 8.1 kPa (IQR 4.9-21.7). 371 patients (69%) had follow-up LSM after a median of 25 months (IQR 17–38), 41 subsequently decompensated and 55 died. Of 125 with cACLD at baseline, 14% were "cACLD increasers" and 43% "cACLD decreasers", while 13% of patients without cACLD were "No cACLD increasers" (n=33/246). Baseline LSM, follow-up LSM and LSM changes accurately predicted decompensation (C-index: Baseline LSM 0.85; Follow-up LSM 0.89; LSM changes 0.85) and mortality (C-index: Baseline LSM 0.74; Follow-up LSM 0.74; LSM changes 0.70). When compared to "cACLD decreasers", "cACLD increasers" had significantly lower decompensation-free survival and higher risks of decompensation (SHR=4.39, P=0.004) and mortality (HR=3.22, P=0.01). Conclusion Liver stiffness by transient elastography predicts decompensation and all-cause mortality in patients with compensated alcohol-related liver disease both at diagnosis and when used for monitoring.