More strokes were observed in the Trial to Reduce Cardiovascular Events With Aranesp Therapy (TREAT) among patients assigned to darbepoetin alfa. We sought to identify baseline characteristics and postrandomization factors that might explain this association.Methods and Results—
A multivariate logistic regression model was used to identify baseline predictors of stroke in 4038 patients with diabetes mellitus, chronic kidney disease, and anemia randomized to receive darbepoetin alfa or placebo. To determine whether postrandomization blood pressure, hemoglobin level, platelet count, or treatment dose were responsible for the increased risk related to darbepoetin alfa, we performed a nested case-control analysis (1:10 matching) identifying nonstroke controls with propensity matching. The risk of stroke was doubled with darbepoetin alfa. Overall, 154 patients had a stroke, 101/2012 (5.0%) in the darbepoetin alfa arm and 53/2026 (2.6%) in the placebo arm (hazard ratio 1.9; 95% confidence interval, 1.4–2.7). Independent predictors of stroke included assignment to darbepoetin alfa (odds ratio 2.1; 95% confidence interval, 1.5–2.9), history of stroke (odds ratio 2.0; 95% confidence interval, 1.4–2.9), more proteinuria, and known cardiovascular disease. In patients assigned to darbepoetin alfa, postrandomization systolic and diastolic blood pressure, hemoglobin level, platelet count, and darbepoetin alfa dose did not differ between those with and without stroke. Additional sensitivity analyses using maximal values, latest values, or changes over varying periods of exposure yielded similar results.Conclusions—
The 2-fold increase in stroke with darbepoetin alfa in TREAT could not be attributed to any baseline characteristic or to postrandomization blood pressure, hemoglobin, platelet count, or dose of treatment. These readily identifiable factors could not be used to mitigate the risk of darbepoetin alfa–related stroke.Clinical Trial Registration—
http://www.clinicaltrials.gov. Unique identifier: NCT00093015.