Patients undergoing maintenance hemodialysis face annual mortality rates of 15–27%, with cardiovascular causes accounting for more than half of all deaths. Despite advances in dialytic technology, conventional prognostic models derived from the general population systematically underestimate risk in end‑stage renal disease, failing to capture the complexity of uremic pathophysiology that drives this excess mortality: immune dysregulation, chronic inflammation, oxidative stress, vascular calcification, uremic cardiomyopathy, protein‑energy wasting, and accelerated vascular aging.

This narrative review synthesizes current evidence on biomarkers predicting all‑cause and cardiovascular mortality in haemodialysis patients, spanning established markers (albumin, CRP, ferritin, hemoglobin, high‑sensitivity troponins, and NT‑proBNP) to an emerging frontier encompassing IL‑6, PTX3, sST2, galectin‑3, FGF‑23, Klotho, and multi‑omics signatures. We further examine composite neutrophil‑derived indices, adipokines, and mineral metabolism markers as tools for pathway‑specific risk interrogation.

The evidence converges on a clear practical message: combination and dynamics outperform single, static measurements. Multi‑biomarker models consistently outperform individual markers, and longitudinal trajectories carry prognostic information, with divergence between survivors and non‑survivors detectable many months before death, that a single time‑point value cannot replicate. Advances in artificial intelligence and multi‑omics further support this shift toward dynamic, personalized risk stratification, though these approaches remain at an early, largely exploratory stage. Critically, clinical translation remains limited by the lack of dialysis‑specific thresholds, poor standardization, and the absence of randomized evidence demonstrating that biomarker‑guided strategies improve outcomes. Bridging this gap between prognostic insight and clinical application remains the central challenge of precision nephrology.