Exploring the relationship between iron metabolism biomarkers and 28 day prognosis in patients with severe acute coronary syndrome: a multicenter retrospective cohort study

Abstract

Iron is an essential trace element for life, playing critical roles in oxygen transport, electron transfer, and DNA synthesis. Recently, it has been increasingly linked to cardiovascular diseases. To investigate this connection, our study analyzed two major clinical databases, MIMIC and eICU, to assess the short-term (28-day) prognostic value of iron metabolism biomarkers—including serum iron, ferritin, transferrin, and total iron-binding capacity (TIBC)—in patients with acute coronary syndrome (ACS). We conducted a retrospective cohort study using data from patients with ACS in the Medical Information Mart for Intensive Care (MIMIC-IV v2.2) database as the internal cohort. For external validation, we utilized the eICU Collaborative Research Database (v2.0). To examine the association between iron biomarkers (serum iron, ferritin, transferrin, and TIBC) and short-term adverse outcomes in ACS patients, we employed multivariate Cox regression, restricted cubic spline (RCS) analysis, and Kaplan-Meier (KM) survival curves. Subsequently, we randomly split the internal cohort into a training set (70%) and a testing set (30%). Univariate logistic regression was applied to identify common prognostic variables across the training set, testing set, and the external validation cohort. Using these shared variables, we developed a multivariate logistic regression model for risk prediction. Finally, the performance of this predictive model was evaluated using receiver operating characteristic (ROC) curve analysis. The study ultimately included 1,652 ACS patients from the MIMIC cohort and 701 from the eICU cohort, with 28-day mortality rates of 10.90% and 20.40%, respectively. In the fully adjusted multivariable model, iron biomarkers showed significant associations with 28-day mortality. Elevated ferritin predicted higher mortality risk (per-unit increase: HR 1.001, 95% CI 1.001–1.002, P = 0.001; Q4 vs. Q1: HR 1.983, 95% CI 1.143–3.440, P = 0.015), while higher transferrin and TIBC were protective. The quartile-based analysis revealed clinically meaningful risk gradients. In contrast, higher levels of transferrin (per-unit increase: HR 0.996, 95% CI 0.993–0.999, P = 0.001; Q4 vs. Q1: HR 0.448, 95% CI 0.265–0.756, P = 0.003) and TIBC (per-unit increase: HR 0.995, 95% CI 0.993–0.998, P = 0.001; Q4 vs. Q1: HR 0.437, 95% CI 0.260–0.734, P = 0.002) were associated with a protective effect. RCS analysis revealed a linear relationship between transferrin, TIBC, and the risk of short-term adverse outcomes, while the relationship for ferritin was nonlinear. Finally, ROC analysis demonstrated that the risk prediction model, which integrated iron biomarkers with eight other independent predictors, outperformed traditional critical illness scoring systems in identifying high-risk ACS patients. The AUC values were 0.70 for the training set, 0.75 for the testing set, and 0.69 for the external eICU validation set. All findings were consistently validated in the external eICU cohort. This study confirms that iron metabolism biomarkers—specifically ferritin, transferrin, and TIBC—serve as independent predictors of 28-day mortality risk in patients with acute coronary syndrome, whereas serum iron showed no significant association with short-term mortality. The risk prediction model developed from these markers demonstrated superior discriminative performance (AUC: 0.69–0.75) compared to conventional scoring systems during both internal and external validation. These biomarkers not only showed statistically significant continuous associations but also enabled clear risk stratification, with extreme quartiles demonstrating markedly different mortality risks. These findings suggest that assessing the iron metabolic profile could aid in the early identification of high-risk patients and provides a valuable foundation for future research into targeted therapeutic strategies.

The authors' abstract, as published at the source. Scientific Reports, 2026 · DOI ↗