Blood samples were collected from 883 individuals in the Uppsala Longitudinal Study of Adult Men (ULSAM) and 978 individuals in the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study. Participants were followed for 10–13 years with n = 113 incident AF cases in ULSAM and n = 148 in PIVUS. The associations between biomarkers and incident AF were analysed in Cox proportional hazards regression models.
The hazard ratio (HR) for incident AF was significant for all five biomarkers in unadjusted analyses in both cohorts. Only NT-proBNP remained significant when adjusting for cardiovascular risk factors and the other biomarkers (HR (1SD) 2.05 (1.62–2.59) (ULSAM) and 1.56 (1.30–1.86) (PIVUS), both p < 0.001). The C-index improved from 0.64 to 0.69 in ULSAM and from 0.62 to 0.68 in PIVUS, by adding NT-proBNP to cardiovascular risk factors (both p < 0.001). The C-index of the CHARGE-AF risk score increased from 0.62 to 0.68 (ULSAM) and 0.60 to 0.66 (PIVUS) by addition of NT-proBNP (p < 0.001).
Using a multimarker approach NT-proBNP was the strongest predictor of incident AF in two cohorts, and improved risk prediction when added to traditional risk factors. NT-proBNP significantly improved the predictive ability of the novel CHARGE-AF risk score, although the predictive value remained modest.