Accumulating data suggest blood biomarkers could inform stroke etiology.
We investigated the performance of multiple blood biomarkers to elucidate stroke etiology with a focus on new-onset atrial fibrillation (AF) and cardioembolism.
Between January and December 2017, information on clinical, laboratory parameters and stroke characteristics were prospectively collected from ischemic stroke patients recruited from the National University Hospital, Singapore. Multiple blood biomarkers (NT-pro-brain derived peptides [NT-proBNP], D-dimer, S100β, neuron-specific enolase [NSE], vitamin D, cortisol, interleukin-6, insulin, uric acid and albumin) were measured in plasma. These variables were compared with stroke etiology, and the risk of new-onset AF and cardioembolism, using multivariable regression methods.
From 515 ischemic stroke patients (mean age, 61 years; 71% men), 44 (8.5%) patients were diagnosed with new-onset AF and 75 (14.5%) patients had cardioembolism. The combination of two laboratory parameters (total cholesterol ≤169 mg/dl and triglycerides ≤44.5 mg/dl) and three biomarkers (NT-proBNP ≥294 pg/ml, S100β ≥64 pg/ml and cortisol ≥471 nmol/l) identified patients with new-onset AF (negative predictive value [NPV] 90%, positive predictive values [PPV] 73% and area under curve [AUC] 85%). The combination of two laboratory parameters (total cholesterol ≤169 mg/dl and triglycerides ≤44.5 mg/dl) and two biomarkers (NT-proBNP ≥507 pg/ml and S100β ≥65 pg/ml) identified those with cardioembolism (NPV 86%, PPV 78% and AUC 87%). Adding clinical predictors, however, did not improve the performance of these models.
Blood biomarkers could identify patients with increased likelihood of cardioembolism and direct the search for occult atrial fibrillation.