BACKGROUND:Higher plasma fibroblast growth factor 21 (FGF21) levels predict incident cardiovascular events in type 2 diabetes patients. However, whether FGF21 levels predict cardiovascular events in statin-treated patients in the general population is unknown. We investigated whether FGF21 levels predict major cardiovascular event (MCVE) in the Treating to New Targets (TNT) trial participants. METHODS:After 8-week run-in on atorvastatin 10 mg/day, 10,001 patients with stable coronary disease in the TNT trial were randomized to 10 mg or 80 mg/day of atorvastatin for a median of 4.9 years. We analyzed data from 1996 patients with plasma FGF21 levels measured at randomization. Among them, 1835 patients had FGF21 measured one-year post-randomization. RESULTS:Higher ln-transformed FGF21 levels at randomization were associated with higher risk of incident MCVE (adjusted hazards ratio per SD increase = 1.18, P = 0.019). At 1-year post-randomization, FGF21 levels were lower in patients randomized to receive 80 mg versus 10 mg atorvastatin (186.9 versus 207.5 pg/mL respectively, P = 0.006). Higher ln-transformed FGF21 levels at 1-year post-randomization were also associated with higher subsequent risk of MCVEs (adjusted hazards ratio per SD increase = 1.24, P = 0.009). However, changes in FGF21 levels over 1-year were not related to subsequent MCVE risk. FGF21 levels had significant incremental value in net reclassification improvement in MCVE risk prediction. CONCLUSIONS:Higher plasma FGF21 levels are associated with higher CVD risk in statin-treated high-risk patients. Higher dose atorvastatin is associated with a reduction in FGF21 levels. FGF21 provides incremental value in CVD risk prediction in statin-treated patients.