Frontiers in Endocrinology · Oct 15, 2026

Serial fibrosis-4 variability and longitudinal patterns are associated with incident heart failure in type 2 diabetes: a 24-month landmark cohort study

By Nagy AC, Ghanem AS, Ulambayar B et al.

BACKGROUND: The prognostic relevance of serial Fibrosis-4 index (FIB-4) changes for incident heart failure (HF) in type 2 diabetes mellitus (T2DM) remains unclear. We examined whether within-person FIB-4 variability and longitudinal patterns were associated with HF after explicit adjustment for chronological age. METHODS: Routinely collected clinical data from 2000 to 2025 were analyzed using a 24-month landmark design. FIB-4 measurements from months 0 to 18 defined the exposure, followed by a 6-month lag before HF follow-up. Variability was calculated as the within-person standard deviation of log₂-transformed FIB-4. Using a threshold of 2.67, patients were classified as never high, improving from high, worsening to high, or persistently high. Associations with incident HF were estimated using multivariable Cox regression. RESULTS: Among 8,221 patients, 1,811 developed HF during 50,640 person-years. Greater FIB-4 variability was associated with a higher HF hazard (adjusted hazard ratio [HR], 1.73; 95% confidence interval [CI], 1.43-2.09). Compared with the never-high pattern, persistent-high FIB-4 (HR, 1.73; 95% CI, 1.35-2.20) and worsening to high (HR, 1.28; 95% CI, 1.01-1.62) were associated with higher HF hazards, whereas improving from high was not (HR, 0.95; 95% CI, 0.70-1.28). Adding serial FIB-4 features resulted in a small, statistically uncertain improvement in temporal-validation discrimination (ΔC-index, 0.0069; 95% CI, -0.0043 to 0.0190). DISCUSSION: Serial FIB-4 variability and persistent elevation were associated with incident HF independently of chronological age. These measures may characterize longitudinal multisystem cardiometabolic burden, but their incremental predictive value was limited. External validation is required before clinical application.

doi.org/10.3389/fendo.2026.1963810 Open access PubMed 42959020 Biomarkers