N-terminal pro-brain natriuretic peptide shows diagnostic accuracy for detecting atrial fibrillation in cryptogenic stroke patients

Background Diagnosing paroxysmal atrial fibrillation in patients with stroke can be difficult. We aimed to determine if N-terminal pro-brain natriuretic peptide can help identify paroxysmal atrial fibrillation in cryptogenic stroke. Methods and results Among 264 ischemic stroke patients, serum levels of N-terminal pro-brain natriuretic peptide were measured within 72 h of stroke onset. In cryptogenic stroke patients, 24-h Holter monitoring was used to look for paroxysmal atrial fibrillation within the first week and also three- and six-months after admission. First, patients with a defined etiology were used to construct a receiver operating characteristic curve for the diagnosis of atrial fibrillation. From this curve, the sensitivity and specificity of preestablished cutoff points for the diagnosis of atrial fibrillation were calculated. A logistic regression was performed to assess the independent relationship of the logarithm of N-terminal pro-brain natriuretic peptide levels with atrial fibrillation. The cutoff points were then evaluated in patients with cryptogenic stroke. Results One hundred eighty-four patients had a specific stroke etiology. Fifty-five patients had atrial fibrillation. Using multivariate analysis, the logarithm of N-terminal pro-brain natriuretic peptide levels was independently associated with atrial fibrillation. The area under the receiver operating characteristic curve of N-terminal pro-brain natriuretic peptide for the diagnosis of atrial fibrillation was 0 center dot 91 (95% confidence interval 0 center dot 87-0 center dot 95). The cutoff point of 265 center dot 5 pg/ml had a sensitivity of 100% and specificity of 70 center dot 5% for the diagnosis of atrial fibrillation. The cutoff point of 912 pg/ml had a sensitivity of 81 center dot 8% and a specificity of 87 center dot 5%. Eighty patients had a cryptogenic stroke. In 17, paroxysmal atrial fibrillation was found during follow-up. In these patients, the area under the curve for the diagnosis of paroxysmal atrial fibrillation was 0 center dot 83. The cutoff point of 265 center dot 5 had a sensitivity of 88 center dot 2% and a specificity of 61 center dot 9%. The cutoff point of 912 pg/ml had a sensitivity of 47 center dot 1% and a specificity of 88 center dot 9%. Conclusion N-terminal pro-brain natriuretic peptide has good accuracy in predicting the presence of paroxysmal atrial fibrillation in patients with cryptogenic stroke and can help to identify these patients.