Jin · Clinical cardiology 2026 · Systematic review and meta-analysis · n=3,724 participants (10 studies)

Circulating Level of Growth-Differentiation Factor 15 and Mortality of Patients With Acute Heart Failure: A Meta-Analysis.

Level 3 - non-randomized controlled study

Meta-analysis of prospective and retrospective observational cohort studies and post-hoc trial analyses.

PubMed 42089382 · doi:10.1002/clc.70338 · record verified 2026-08-26

What was done

Authors systematically searched PubMed, Embase, and Web of Science for prospective or retrospective cohort studies and post-hoc trial analyses that evaluated circulating growth differentiation factor-15 (GDF-15) measured at admission in adults hospitalized with acute heart failure (AHF). Risk ratios (RRs) for all-cause mortality comparing high versus low GDF-15 levels were pooled using random-effects meta-analysis.

What was found

Ten studies enrolling 3,724 patients met inclusion criteria. High admission GDF-15 was significantly associated with increased follow-up mortality (RR = 2.82, 95% CI: 2.39–3.32; p < 0.001) with no detectable between-study heterogeneity (I² = 0%). The association remained consistent across leave-one-out sensitivity analyses (RR range: 2.73–3.00) and in high-quality studies scoring Newcastle-Ottawa Scale ≥8 (RR = 2.72, 95% CI: 2.26–3.27). Subgroup comparisons showed no significant differences by cohort region (Asian vs. Western), study design, sampling timing (admission to 48 h), assay method (ELISA vs. ECLIA), cutoff definition, follow-up duration, or adjustment for BNP/NT-proBNP (all subgroup p > 0.05). Publication bias was not detected (Egger's p = 0.59).

Why it matters

Elevated GDF-15 at admission provides strong, consistent prognostic information for mortality in acute heart failure independent of natriuretic peptides, supporting its potential role in early risk stratification.

Limits

The analysis relies entirely on observational cohort data and post-hoc trial subsets, precluding causal inference. Cutoff definitions, assay platforms, and exact follow-up durations varied across included studies rather than adhering to a single standardized clinical threshold. The abstract does not provide absolute risk differences or measures of incremental prognostic utility (such as C-statistic improvements).

Cited by