NT-proBNP and UACR for outcome prediction in HFrEF

Background/Introduction

NTproBNP is the central biomarker for risk stratification in heart failure with reduced ejection fraction (HFrEF). The urinary albumin-to-creatinine ratio (UACR) reflects renal and microvascular injury, but its independent and incremental prognostic value beyond NTproBNP remains uncertain.

Purpose

To evaluate the prognostic relevance of UACR in chronic HFrEF and its additive value beyond NTproBNP.

Methods

We analyzed 867 patients with chronic HFrEF with baseline NTproBNP and UACR. Outcomes were heart failure hospitalization (HFH), all-cause death, HFH or death, and a combined renal endpoint (hospitalization for acute kidney injury, progression to kidney failure, or initiation of renal replacement therapy). Cox regression models with interaction terms, ROC/AUC analysis, and Kaplan-Meier analyses were performed.

Results

50% of patients had UACR <30 mg/g. Higher UACR was associated with ischemic etiology, higher NTproBNP tertiles, and advanced NYHA class (Figure 1, p for all <0.05).

In univariable Cox models, UACR and NTproBNP were significant predictors across endpoints. For HFH or death, UACR (HR: 1.37 [1.23–1.53]; p<0.001) and NTproBNP (HR: 2.04 [1.77–2.35]; p<0.001) were both associated with risk; in combined models, NTproBNP remained independently predictive (HR: 2.06 [1.75–2.43]; p<0.001), whereas UACR was attenuated (HR: 1.15 [0.98–1.36]; p=0.086). For all-cause death, UACR lost significance after adjustment, while NTproBNP remained associated (adjusted HR: 2.20 [1.70–2.85]; p<0.001). For HFH, NTproBNP remained independently predictive, whereas UACR showed independent association (adjusted HR: 1.22 [1.01–1.47]; p=0.038). For the combined renal endpoint, UACR remained independently associated even in models including NTproBNP (HR: 1.50 [1.07–2.09]; p=0.018).

In ROC analyses, NTproBNP consistently outperformed UACR (e.g., death AUC: 0.724 [0.681–0.768] vs. 0.628 [0.578–0.677]; p<0.001), and adding UACR did not significantly improve discrimination beyond NTproBNP.

There was significant interaction between UACR and NTproBNP for HFH or death (p=0.03) and for the combined renal endpoint (p=0.018).

Kaplan-Meier analyses using clinically defined cutoffs (NTproBNP 1000pg/mL; UACR 30mg/g) demonstrated clear separation across all four endpoints, with the highest event rates observed in patients with both elevated biomarkers (Figure 2, log-rank p<0.001 for all).

Conclusion

UACR is associated with adverse HF and renal outcomes in chronic HFrEF but does not provide consistent independent prognostic value beyond NTproBNP. However, UACR seems to refine risk primarily in patients with elevated NTproBNP, but not when NTproBNP is low. A simple dual-threshold approach using NTproBNP 1000pg/mL and UACR 30mg/g identifies a distinctly high-risk subgroup across all HF and renal endpoints, supporting a pragmatic and clinically intuitive strategy for risk stratification.