- Design
- Meta-analysis and meta-regression of RCTs and propensity-matched cohorts
- Population
- 1958 patients with acute-on-chronic liver failure in 11 studies
- Primary outcome
- 28-day mortality
- Effect
- RR 0.72 (95% CI 0.63–0.83); MELD 25–30 RR 0.65 (0.52–0.81)
A meta-analysis pooled 11 studies (12 datasets, 1958 patients) comparing non-bioartificial liver support — techniques such as plasma exchange and adsorption — with standard medical therapy in acute-on-chronic liver failure. Studies were randomised trials or propensity-matched cohorts.
Liver support was associated with lower 28-day mortality (RR 0.72, 95% CI 0.63–0.83) with little heterogeneity. The benefit was clearest at MELD 25–30 (RR 0.65) and present above 30 (RR 0.75); below 25, the estimate was too imprecise to interpret. Meta-regression placed the largest benefit at a baseline MELD of about 25–32.
Plasma exchange is widely used for ACLF in Indian centres, often without clear selection rules. This analysis suggests patient selection by severity matters. It mixes randomised and observational data, so the MELD window is a hypothesis for trials rather than a validated threshold.
- Consider liver support for ACLF within a structured protocol, ideally at a transplant-capable centre.
- The clearest signal of benefit was in patients with MELD about 25–32.
- Evidence in MELD below 25 is too uncertain to guide practice.
- Liver support is a bridge, not a substitute for transplant assessment.
Why it matters
A procedure already common in Indian liver units may help some patients more than others.
Don't overread it
The MELD window comes from subgroup analysis and meta-regression mixing trials with cohorts; it is not a validated threshold.
The statistics, in plain English
A risk ratio of 0.72 means about 28% fewer deaths at 28 days. Subgroup results in meta-analysis are weaker than the overall estimate because they compare across different studies.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for gastroenterology & hepatology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free