SAA (Serum Amyloid A): A Novel Predictor of Stroke-Associated Infections

• Published in: Stroke (1970) 2020-12, Vol.51 (12), p.3523-3530
• Main Authors: Schweizer, Juliane, Bustamante, Alejandro, Lapierre-Fétaud, Vanessa, Faura, Júlia, Scherrer, Natalie, Azurmendi Gil, Leire, Fluri, Felix, Schütz, Valerie, Luft, Andreas, Boned, Sandra, Sanchez, Jean-Charles, Montaner, Joan, Katan, Mira
• Format: Article
• Language: English
• Subjects: Aged; Aged, 80 and over; Area Under Curve; Biomarkers; C-Reactive Protein - metabolism; Clinical Decision Rules; Cross Infection - epidemiology; Cross Infection - metabolism; Deglutition Disorders - physiopathology; Female; Healthcare-Associated Pneumonia - epidemiology; Healthcare-Associated Pneumonia - metabolism; Humans; Ischemic Stroke - metabolism; Ischemic Stroke - physiopathology; Ischemic Stroke - therapy; Leukocyte Count; Logistic Models; Male; Middle Aged; Procalcitonin - metabolism; Reproducibility of Results; ROC Curve; Sepsis - metabolism; Sepsis - physiopathology; Sepsis - therapy; Serum Amyloid A Protein - metabolism; Urinary Tract Infections - metabolism; Urinary Tract Infections - physiopathology; Urinary Tract Infections - therapy
• ISSN: 0039-2499; 1524-4628
• DOI: 10.1161/STROKEAHA.120.030064

BACKGROUND AND PURPOSE:The aim of this study was to evaluate and independently validate SAA (serum amyloid A)—a recently discovered blood biomarker—to predict poststroke infections. METHODS:The derivation cohort (A) was composed of 283 acute ischemic stroke patients and the independent validation cohort (B), of 367 patients. The primary outcome measure was any stroke-associated infection, defined by the criteria of the US Centers for Disease Control and Prevention, occurring during hospitalization. To determine the association of SAA levels on admission with the development of infections, logistic regression models were calculated. The discriminatory ability of SAA was assessed, by calculating the area under the receiver operating characteristic curve. RESULTS:After adjusting for all predictors that were significantly associated with any infection in the univariate analysis, SAA remained an independent predictor in study A (adjusted odds ratio, 1.44 [95% CI, 1.16–1.79]; P=0.001) and in study B (adjusted odds ratio, 1.52 [1.05–2.22]; P=0.028). Adding SAA to the best regression model without the biomarker, the discriminatory accuracy improved from 0.76 (0.69–0.83) to 0.79 (0.72–0.86; P