BNP over NT-proBNP to predict incident atrial fibrillation after cryptogenic stroke.

BACKGROUND AND PURPOSE BNP and NT-proBNP are well known surrogates of atrial fibrillation (AF) detection but studies usually present data on either BNP or NT-proBNP. We aimed to determine and directly compare the validity of both biomarkers as a tool to predict AF and guide prolonged cardiac monitoring in cryptogenic stroke patients. METHODS Non-lacunar acute ischemic stroke (<72h) patients over 55 years of age, with cryptogenic stroke after standard evaluation were included in the Crypto-AF study and blood was collected. BNP and NT-proBNP levels were determined by automated immunoassays. AF was assessed by 28 days monitoring. Highest (optimizing specificity) and lowest quartiles (optimizing sensitivity) were used as biomarker cut-offs to build predictive models adjusted by sex and age. The integrated discrimination improvement index (IDI) and DeLong test were used to compare the performance of both biomarkers. RESULTS From 320 patients evaluated, 218 were included in the analysis. AF was detected in 50 patients (22.9%). NT-proBNP (p<0.001) and BNP (p<0.001) levels were higher in subjects with AF and their levels correlated (r=0.495, p<0.001). BNP showed an increased area under the curve (0.720 vs 0.669; p=0.0218) and a better predictive capacity (IDI=3.63% [1.36%-5.91%]) compared to NT-proBNP. BNP performed better than NT-proBNP in a specific model (IDI=3.7%[0.87-6.5]), while both biomarkers performed similar in the case of a sensitive model. CONCLUSIONS Both BNP and NT-proBNP were increased in cryptogenic stroke patients with AF detection. Interestingly, BNP outperforms NT-proBNP, especially in terms of specificity.