Abstract
OBJECTIVE: To examine fibroblast growth factor 23 (FGF23) as a risk factor for incident stroke in a racially/ethnically diverse population-based urban cohort. METHODS: Stroke-free Northern Manhattan Study participants with FGF23 measurements (n = 2,525) were followed for a mean of 12 (±5) years to detect incident strokes. We used Cox proportional hazards models to estimate the association of baseline FGF23 with incident total, ischemic, and hemorrhagic stroke. RESULTS: Median FGF23 was 57 relative units (RU)/mL (interquartile range = 44-81 RU/mL). Each unit increase of natural log-transformed FGF23 conferred a 40% greater overall stroke risk after adjusting for estimated glomerular filtration rate and sociodemographic and vascular risk factors (hazard ratio = 1.4, 95% confidence interval 1.1-1.6, p = 0.004). Penalized spline analysis revealed a linear association with overall stroke risk at ≥90 RU/mL FGF23, compared with <90 RU/mL (hazard ratio = 1.5, 95% confidence interval = 1.2-2.1, p = 0.004). Greater FGF23 conferred a doubling of intracerebral hemorrhage (ICH) risk but no significant increased risk of ischemic stroke. The associations of elevated FGF23 levels with greater risks of overall stroke and ICH events were independent of phosphate and parathyroid hormone levels and were similar among participants without chronic kidney disease. CONCLUSIONS: Elevated FGF23 was a risk factor for overall stroke and ICH events, in particular in a racially and ethnically diverse urban community, independent of chronic kidney disease.