Brain natriuretic peptide changes as a potential biomarker for paroxysmal atrial fibrillation in acute ischemic stroke cases

BACKGROUND AND PURPOSE: Identifying reliable predictors for paroxysmal atrial fibrillation (PAF) in acute ischemic stroke (AIS) remains a critical clinical challenge. Since conversion from AF to sinus rhythm reduces plasma brain natriuretic peptide (BNP) levels, we hypothesized that BNP reduction could serve as a biomarker for detecting PAF in AIS cases. METHODS: We performed a retrospective analysis of consecutive AIS cases admitted to our institution from 2022 to 2025, measuring BNP twice: upon admission and within 48 hours. Changes in BNP levels (ΔBNP) and logarithmically transformed BNP values (ΔLog BNP), and multiple stroke-associated covariates were compared across three groups: cardioembolic stroke due to PAF (CES-PAF), other CES etiologies (CES), and non-CES. The diagnostic performance of ΔLog BNP for CES-PAF was evaluated and compared with established PAF predictors. RESULTS: Among 400 consecutive AIS cases, 220 were eligible. Median ΔBNP (pg/mL) was -19.6 in the CES-PAF group (n = 30), +82.1 in the CES group (n = 67), and +25.5 in the non-CES group (n = 123). ΔLog BNP was significantly lower in the CES-PAF group (vs. CES: p < 0.001, vs. non-CES: p < 0.001). This association remained significant following multivariate adjustment for stroke-associated covariates (vs. CES: p < 0.001, vs. non-CES: p < 0.001). ΔLog BNP demonstrated superior predictive performance compared with established PAF predictors, with an area under the ROC curve of 0.91 (95 % CI: 0.84-0.98). Composite scores incorporating ΔLog BNP further enhanced diagnostic performance. CONCLUSIONS: These findings suggest that ΔBNP could be a promising practical clinical biomarker for identifying PAF in patients with causally related AIS, warranting validation in larger prospective cohorts.