NT-proBNP Thresholds for Early Heart Failure Detection in Asian Patients With Type 2 Diabetes

Abstract Background Heart failure (HF) is an increasingly common complication among patients with type 2 diabetes (T2D), yet its early detection remains challenging, especially in those with concomitant chronic kidney disease (CKD). NT-proBNP is a key biomarker for diagnosing and prognosticating HF, but its reference thresholds are influenced by renal function, age, and ethnicity. Current guideline cutoffs, largely derived from Western populations, may not apply to Asian patients. Methods This retrospective cohort study included 10,587 adults with T2D who underwent NT-proBNP testing between 2006 and 2021 at the National Taiwan University Hospital. Patients with prior HF were excluded. Generalized additive models identified NT-proBNP thresholds associated with HF hospitalization, and Kaplan–Meier analysis validated outcome separation. Subgroup analyses were stratified by age, sex, body mass index (BMI), and estimated glomerular filtration rate (eGFR). Results During a mean follow-up of 3.5 years, 1,892 (17.9%) patients were hospitalized for HF. NT-proBNP levels of 179 pg/mL (outpatient) and 728 pg/mL (emergency) marked inflection points for rising event risk (log-rank p < 0.0001). Age-specific analyses showed progressive increases in optimal thresholds: from 85 ( 60, 30–60, and < 30 mL/min/1.73 m², respectively; in the emergency cohort, the corresponding thresholds were 290, 835, and 3,905 pg/mL. Conclusions This large Asian cohort defines setting- and renal function-specific NT-proBNP thresholds for predicting HF hospitalization in patients with T2D. The lower optimal cutoffs compared with Western guidelines highlight the need for ethnicity-adjusted diagnostic criteria to improve early identification and risk stratification of HF in clinical practice. What is new? In a large real-world Asian cohort of patients with type 2 diabetes, we identified setting-specific NT-proBNP thresholds (179 pg/mL outpatient; 728 pg/mL emergency) associated with heart failure hospitalization risk. Age-, BMI-, and kidney function–stratified cutoffs revealed substantial heterogeneity in optimal NT-proBNP thresholds. Compared with guideline-recommended values, Asian-specific thresholds were consistently lower (∼30–40%), supporting ethnic differences in natriuretic peptide biology. A generalized additive model (GAM) captured nonlinear biomarker–risk relationships, enabling data-driven and clinically interpretable cutoff identification. What are the clinical implications? Use of ethnicity- and context-specific NT-proBNP thresholds may improve early detection of heart failure in Asian patients with type 2 diabetes. Incorporating kidney function and BMI into NT-proBNP interpretation enhances risk stratification, particularly in patients with CKD. Reliance on Western guideline cutoffs may underestimate heart failure risk in Asian populations. These findings support a precision medicine approach to biomarker interpretation and highlight the need for population-specific guideline refinement.