A total of 1,668 patients were initially evaluated for inclusion. After excluding those with missing data for any study variable, 907 AHF patients were retained in the final analysis [135 (15%) from Japan and 772 (85%) from Europe]. Japanese patients were enrolled at Nippon Medical School Musashikosugi Hospital (Kawasaki, Japan) and Kawaguchi Cardiovascular and Respiratory Hospital (Kawaguchi, Japan), while European patients were enrolled from two pre-registered prospective studies: the Lithuanian Echocardiography Study of Dyspnoea in Acute Settings (NCT03048032) and Diagnostic and Prognostic Value of New Biomarkers in Patients With Heart Disease study (NCT01374880), conducted at multiple European sites. The study was approved by local ethics committees: Lithuanian Bioethics Committee (No. L-15-01), Comité d’Evaluation de l’Ethique des Projets de Recherche Biomédicale (CEERB) (No. 10-017), Nippon Medical School Musashi-Kosugi Hospital (No. 220-24-20) and Kawaguchi Cardiovascular and Respiratory Hospital (No. 2016-003). Informed consent was obtained from all patients.

AHF was defined according to European Society of Cardiology guidelines [1]. Clinical and biological data were collected upon admission into the emergency department for acute dyspnea by reviewing electronic medical records. Blood samples were collected within 4 hrs of presentation, stored at –80°C, and transferred to INSERM UMR-S 942 (Paris, France) for centralized biomarker analyses, including single baseline measurements of NT-proBNP, high-sensitivity troponin-T (hs-TnT), growth differentiation factor 15 (GDF-15), interleukin-6 (IL-6), and C-reactive protein (CRP) levels, using commercial assays (Roche Diagnostics, Mannheim, Germany, Europe). NT-proBNP, hs-TnT, GDF-15, and IL-6 levels were measured using electrochemiluminescence immunoassays on a Cobas E801 analyzer (Roche Diagnostics, Meylan, France). CRP concentrations were measured using the Tina-quant CRP-Gen.3 immunoturbidimetric assay on a Cobas c701 analyzer (Roche Diagnostics). Analytical performance metrics (including limits of detection and measurement ranges) were provided by the manufacturer. CVs, obtained through internal QC testing during the study, were <5% and in accordance with our laboratory standards.

Baseline characteristics were compared using Pearson’s chi-squared or Fisher’s exact test for categorical variables and Wilcoxon rank-sum test for continuous or ordinal variables, as appropriate. Biomarkers levels were compared between groups using both the Wilcoxon rank-sum test and standardized mean differences (SMDs) with 95% confidence intervals. An absolute SMD <0.1 was considered indicative of adequate balance. Propensity score matching (PSM) was conducted using logistic regression to estimate scores based on age, sex, body mass index (BMI), left ventricular ejection fraction (LVEF), systolic blood pressure, heart rate, estimated glomerular filtration rate, history of diabetes mellitus, chronic HF, atrial fibrillation, and coronary artery disease. Patients were matched in a 3:1 ratio via nearest-neighbor matching with a caliper of 0.8 SDs of the logit of the propensity score. For sensitivity analysis, multivariable logistic regression was used to assess the association between ethnicity and NT-proBNP levels after matching, adjusting for renal function (serum creatinine), age, and BMI, which are known clinical confounders influencing NT-proBNP levels.

Several differences were noted prior to PSM. Japanese patients had significantly lower Hb levels [119 g/L (IQR 100.5, 134.0) vs. 130 g/L (IQR 115.0, 143.0), P<0.001]. Median LVEF was 40% (IQR 20, 54) in European patients and 45% (IQR 32, 60) in Japanese patients (P<0.001). Japanese patients were less frequently treated with standard HF medications at discharge [beta-blockers: 12 (8.9%) vs. 450 (59%), P<0.001, renin-angiotensin antagonists: 52 (39%) vs. 395 (52%), P=0.02]. After 3:1 PSM, differences became either non-significant or were markedly reduced (Table 1, Supplemental Data Table S1, and Supplemental Data Fig. S1).

NT-proBNP levels were significantly higher in Japanese AHF patients than in their European counterparts [4,060 ng/L (IQR 2,081, 12,218) vs. 3,390 ng/L (IQR 1,410, 7,682), P=0.004] (Table 2). After PSM, NT-proBNP levels showed no significant difference between the two groups [4,040 ng/L (IQR 2,084, 11,356) in Japan vs. 3,722 ng/L (IQR 1,324, 8,727), P=0.083], with a SMD of –0.09, suggesting an adequate balance between the cohorts. After stratification according to LVEF at admission (<40%, 40%–50%, and >50%), NT-proBNP levels were significantly higher in Japanese AHF patients, particularly in the group with LVEF >50% (N=118 vs. 47, P=0.02) (Fig. 1A), with no significant differences in the <40% (N=208 vs. 45, P=0.97) and 40%–50% subgroups (N=100 vs. 29, P=0.11). However, post-PSM NT-proBNP levels were not significantly different in any LVEF subgroup (<40%, N=55 vs. 36, P=0.79; 40%–50%, N=137 vs. 26, P=0.21; >50%, N=51 vs. 34, P=0.36) (Fig. 1B).

hs-TnT, GDF-15, IL-6, and CRP levels did not significantly differ between Japanese and European AHF patients at baseline or after PSM. The sensitivity analysis revealed that, after adjusting for renal function (serum creatinine), age, and BMI, the NT-proBNP levels did not differ significantly between Japanese and European patients (estimate=0.935, P=0.666; Supplemental Data Table S2).

This study demonstrated that biomarker levels, particularly NT-proBNP levels, do not differ significantly between European and Japanese AHF patients. Analyses based on LVEF subclasses also did not reveal significant differences. However, importantly, our study was underpowered to conclusively detect differences among the LVEF subgroups.

Precipitating factors in AHF show ethnic differences [5]; for instance, anemia, primarily caused by iron deficiency, is more prevalent in the Japanese population [6], as also observed in our cohort. Iron deficiency affects up to 80% of AHF patients and may act as a triggering factor [1].

As for HF medications, the higher rate of beta-blocker use at admission in European patients likely reflects a greater prevalence of chronic HF in this group, as opposed to the more frequent de novo presentations in Japanese patients. However, this difference is unlikely to have significantly influenced NT-proBNP levels, particularly given the absence of between-group differences in diuretic prescription at admission after matching, suggesting comparable disease severity at presentation. Regarding medications for HF at discharge, the observed differences should not be interpreted in the context of different clinical guidelines. We analyzed data collected between 2011 and 2017, when older HF guidelines comparable between regions were in use. Hence, the observed differences in medical therapy cannot be attributed to variations in treatment indications. European patients better adhered to the guidelines from that period, as opposed to the suboptimal management among Japanese HF patients. Although data on sodium glucose cotransporter 2 inhibitor (SGLT2i) and angiotensin receptor/neprilysin inhibitor (ARNI) use were not available for our cohorts, these treatments represent key components of modern HF management and may influence NT-proBNP levels. NT-proBNP levels do not typically increase through neprilysin inhibition during ARNI treatment; instead, persistently elevated values are associated with worse outcomes because of ongoing congestion or disease progression [7]. In contrast, SGLT2is lower NT-proBNP levels, likely through diuretic effects exerted both alone and synergistically with other HF therapies and favorable cardiac remodeling [8, 9].

Underutilization of guideline-directed medical therapy (GDMT) has been observed across ethnic groups [10], and achieving adequate prescription and dose titration of GDMT among demographic groups remains a challenge in ensuring quality HF care. A Japanese observational study revealed a reduced use of cardioprotective medications for HF with reduced ejection fraction at discharge [11].

Our study had some limitations. First, this was an observational study, which may have led to selection and confounding biases. Second, in the European database, ethnicities were not specified. While this reflects real-world European clinical practice, with patients of various ethnicities, further large studies from both regional and ethnic perspectives are required. Third, the sample size was small, limiting representativeness and statistical power. Fourth, the absence of healthy control data limited contextualization of biomarker levels within normal ranges. Fifth, the statistical power was insufficient to reliably detect differences within LVEF subgroups, which may affect interpretation of subgroup analyses. Finally, data regarding medical therapy did not reflect current guidelines and clinical practice, although we believe that comparable treatment exposure would not have substantially affected our findings. In this regard, studies using contemporary data will provide more insights.

In conclusion, NT-proBNP levels do not differ significantly between Caucasian and Asian AHF patients with similar clinical presentations, regardless of LVEF. Our results support that NT-proBNP levels in AHF patients can be interpreted without requiring ethnicity-specific adjustments in these populations.