Role of NT-proBNP for Atrial Fibrillation Detection After Ischemic Stroke: A Time-Dependent Relationship

Atrial fibrillation detected after stroke (AFDAS) affects secondary stroke prevention, yet identification can be challenging. Easily accessible cardiac blood biomarkers such as NT-proBNP (N-terminal pro-B-type natriuretic peptide) could guide diagnostic workup, but optimal cutoff values and the time-dependent relationship between NT-proBNP and AFDAS are unclear. We aimed (1) to externally validate earlier presented NT-proBNP cutoffs for atrial fibrillation prediction and (2) to assess the time-dependent relationship of NT-proBNP and early in-hospital AFDAS versus AFDAS after discharge.

We conducted a pooled data analysis of patients with ischemic stroke from the prospective international multicenter BIOSIGNAL (Biomarker Signature of Stroke Aetiology) cohort study (European Stroke Centers from October 2014 to October 2017) and the prospective single-center Graz stroke pathway study (Austria from May 2018 to August 2020). AFDAS was defined as ≥30-s atrial fibrillation/flutter diagnosed within 1 year post-admission and categorized in in-hospital versus after discharge. NT-proBNP was assessed ≤24 hours of symptom onset. The association between NT-proBNP and AFDAS was evaluated by a multivariable logistic regression analysis.

AFDAS was diagnosed in 374 (16%) of 2292 patients with ischemic stroke (median age, 74 years; 42% female), 268 (72%) during hospitalization, and 106 (28%) after discharge (median duration of hospitalization, 15 days). NT-proBNP levels at admission had a good predictive capacity for in-hospital AFDAS (area under the receiver operating characteristic curve, 0.83 [95% CI, 0.81-0.86]). For patients diagnosed with AFDAS after discharge, the predictive capacity of NT-proBNP was poor (area under the receiver operating characteristic curve, 0.65 [95% CI, 0.60-0.70]), and 20% had normal NT-proBNP values <125 pg/mL at admission. The NT-proBNP cutoff of 505 pg/mL exhibited high sensitivity (82%) and specificity (71%) for in-hospital AFDAS, with a negative predictive value of 96%.

In patients with ischemic stroke, the admission NT-proBNP cutoff of 505 pg/mL seems to be a reliable predictor for in-hospital AFDAS, while the predictive capacity of NT-proBNP for AFDAS after discharge is limited. Our results might influence the designs of future secondary stroke prevention trials.